U.S. patent application number 10/326526 was filed with the patent office on 2004-02-19 for displacement measurement method and apparatus, strain measurement method and apparatus elasticity and visco-elasticity constants measurement apparatus, and the elasticity and visco-elasticity constants measurement apparatus-based treatment apparatus.
Invention is credited to Sumi, Chikayoshi.
Application Number | 20040034304 10/326526 |
Document ID | / |
Family ID | 31721996 |
Filed Date | 2004-02-19 |
United States Patent
Application |
20040034304 |
Kind Code |
A1 |
Sumi, Chikayoshi |
February 19, 2004 |
Displacement measurement method and apparatus, strain measurement
method and apparatus elasticity and visco-elasticity constants
measurement apparatus, and the elasticity and visco-elasticity
constants measurement apparatus-based treatment apparatus
Abstract
The present invention provides elastic constant and visco
elastic constant measurement apparatus etc. for measuring in the
ROI in living tissues elastic constants such as shear modulus,
Poisson's ratio, Lame constants, etc., visco elastic constants such
as visco shear modulus, visco Poisson's ratio, visco Lame
constants, etc. and density even if there exist another mechanical
sources and uncontrollable mechanical sources in the object. The
elastic constant and visco elastic constant measurement apparatus
is equipped with means of data storing 2 (storage of deformation
data measured in the ROI 7 etc.) and means of calculating elastic
and visco elastic constants 1 (calculator of shear modulus etc. at
arbitrary point in the ROI from measured strain tensor data etc.),
the means of calculating elastic and visco elastic constants
numerically determines elastic constants etc. from the first order
partial differential equations relating elastic constants etc. and
strain tensor etc.
Inventors: |
Sumi, Chikayoshi;
(Tokorozawa-shi, JP) |
Correspondence
Address: |
SUGHRU MION, PLLC
2100 Pennsylvania Avenue, NW
Washington
DC
20037-3213
US
|
Family ID: |
31721996 |
Appl. No.: |
10/326526 |
Filed: |
December 23, 2002 |
Current U.S.
Class: |
600/439 |
Current CPC
Class: |
G01S 15/8977 20130101;
A61B 8/488 20130101; A61B 5/0051 20130101; G01S 15/8979 20130101;
A61B 5/015 20130101; G01S 7/52042 20130101; G01S 15/899 20130101;
A61B 8/485 20130101; A61B 8/08 20130101; A61N 7/02 20130101; A61B
5/0053 20130101 |
Class at
Publication: |
600/439 |
International
Class: |
A61B 008/00; A61B
007/00 |
Foreign Application Data
Date |
Code |
Application Number |
Dec 21, 2001 |
JP |
2001-389484 |
Jan 18, 2002 |
JP |
2002-10368 |
Apr 25, 2002 |
JP |
JP2002-123765 |
May 17, 2002 |
JP |
JP2002-143145 |
Jun 17, 2002 |
JP |
JP2002-176242 |
Jun 19, 2002 |
JP |
JP2002-178265 |
Aug 19, 2002 |
JP |
JP2002-238434 |
Aug 21, 2002 |
JP |
JP2002-240339 |
Aug 22, 2002 |
JP |
JP2002-241647 |
Sep 2, 2002 |
JP |
JP2002-256807 |
Claims
1. The stated displacement measurement methods for measuring local
displacement vector or local displacement vector components from
the phases of the ultrasound echo signals acquired from the target
as the responses to more than one time transmitted ultrasound.
2. The displacement measurement methods of claim 1 that further
include means of thinning out acquired ultrasound echo signals in
each direction.
3. The displacement measurement methods of claim 1 that further
include means of sine-modulation of transmitted beams' amplitudes
in scan direction.
4. The displacement measurement methods of claim 1 that further
include both means of beam steering and sine-modulation of
transmitted beams' amplitudes in scan direction.
5. The displacement measurement methods of claim 1 that further
include means of utilizing at least one of ultrasound echo basic
wave components and ultrasound echo harmonic wave components as
ultrasound echo signals.
6. The displacement measurement apparatus equipped with the
following means: displacement (strain) sensor (transducer to
transmit ultrasounds to the target, and detect echo signals
generated in the target), relative position controller and relative
direction controller between the sensor and the target, means of
transmitting/receiving (transmitter of driving signals to the
sensor, and receiver of the echo signals detected at the sensor),
means of data processing (controller of the driving signals of the
means of transmitting, and processor of the received echo signals
of means of receiving), and means of data storing (storage of echo
signals, measured deformation data); wherein the means of data
processing also measures the local displacement vector or the local
displacement vector components utilizing the stated displacement
measurement methods from the phases of the ultrasound echo signals
acquired from the target as the responses to more than one time
transmitted ultrasound.
7. The strain tensor measurement apparatus equipped with the
displacement measurement apparatus of claim 6, and the means of
data processing can also yield strain tensor components by spatial
differential filtering with suitable cut off frequency in spatial
domain or frequency domain the measured 3D, or 2D displacement
vector components, or measured 1D direction displacement component
in the 3D, 2D, or 1D ROI, and the means of data processing can also
yield strain rate tensor components, acceleration vector
components, or velocity vector components by time differential
filtering with suitable cut off frequency in time domain or
frequency domain the measured time series of displacement
components, or strain components.
8. The stated strain tensor measurement methods for directly
measuring the local strain tensor or the local strain tensor
components from the phases of the ultrasound echo signals acquired
from the target as the responses to more than one time transmitted
ultrasound.
9. The strain tensor measurement methods of claim 8 that further
include means of sine-modulation of transmitted beams' amplitudes
in scan direction.
10. The strain tensor measurement methods of claim 8 that further
include both means of beam steering and sine-modulation of
transmitted beams' amplitudes in scan direction.
11. The strain tensor measurement methods of claim 8 that further
include means of utilizing at least one of ultrasound echo basic
wave components and ultrasound echo harmonic wave components as
ultrasound echo signals.
12. The strain tensor measurement apparatus equipped with the
following means: displacement (strain) sensor (transducer to
transmit ultrasounds to the target, and detect echo signals
generated in the target), relative position controller and relative
direction controller between the sensor and the target, means of
transmitting/receiving (transmitter of driving signals to the
sensor, and receiver of the echo signals detected at the sensor),
means of data processing (controller of the driving signals of the
means of transmitting, and processor of the received echo signals
of means of receiving), and means of data storing (storage of echo
signals, measured deformation data; wherein the means of data
processing also directly measures the local strain tensor or the
local strain tensor components utilizing the stated direct strain
measurement methods from the phases of the ultrasound echo signals
acquired from the target as the responses to more than one time
transmitted ultrasound.
13. The strain tensor measurement apparatus equipped with the
strain measurement apparatus of claim 12, and the means of data
processing can also yield strain rate tensor components by time
differential filtering with suitable cut off frequency in time
domain or frequency domain the measured time series of 3D, or 2D
strain tensor components, or measured 1D direction strain component
in the 3D, 2D, or 1D ROI.
14. The elasticity and visco-elasticity constants measurement
apparatus equipped with the following means: means of data storing
(storage of at least one of strain tensor data, strain rate tensor
data, acceleration vector data, elastic constants, visco elastic
constants, or density measured in the ROI set in the target), and
means of calculating elastic and visco elastic constants
(calculator of at least one of elastic constants, visco elastic
constants, or density of arbitrary point in the ROI from at least
one of the measured strain tensor data, strain rate tensor data, or
acceleration vector data); wherein the means of calculating elastic
and visco elastic constants numerically determines at least one of
the elastic constants, visco elastic constants, or density from the
first order partial differential equations relating at least one of
the elastic constants, visco elastic constants, or density to at
least one of the strain tensor data, strain rate tensor data,
acceleration vector data.
15. The elasticity and visco-elasticity constants measurement
apparatus equipped with the following means: means of data storing
(storage of at least one of strain tensor data, strain rate tensor
data, acceleration vector data, elastic constants, visco elastic
constants, or density measured in the ROI including lesions), means
of calculating elastic and visco elastic constants (calculator of
at least one of elastic constants, visco elastic constants, or
density of arbitrary point in the ROI from at least one of the
measured strain tensor data, strain rate tensor data, or
acceleration vector data), and means of output of degeneration
information on parts including the lesions (output means of
degeneration information based on calculated at least one of the
elastic constants, visco elastic constants, or density); wherein
the means of calculating elastic and visco elastic constants
numerically determines at least one of the elastic constants, visco
elastic constants, or the density from the first order partial
differential equations relating at least one of the elastic
constants, visco elastic constants, or density to at least one of
the strain tensor data, strain rate tensor data, acceleration
vector data.
16. The elasticity and visco-elasticity constants measurement
apparatus of claim 15, in which means of output outputs
degeneration information by comparing calculated elastic constants,
visco elastic constants, density with set in advance those
values.
17. The elasticity and visco-elasticity constants measurement
apparatus of claim 14, in which means of calculating elastic and
visco elastic constants utilizes at least one of predicted in
advance or set in advance the elastic constants, visco elastic
constants, density as the coefficients of the prescribed first
order partial differential equations.
18. The elasticity and visco-elasticity constants measurement
apparatus of claim 15, in which means of calculating elastic and
visco elastic constants utilizes at least one of predicted in
advance or set in advance the elastic constants, visco elastic
constants, density as the coefficients of the prescribed first
order partial differential equations.
19. The elasticity and visco-elasticity constants measurement
apparatus of claim 16, in which means of calculating elastic and
visco elastic constants utilizes at least one of predicted in
advance or set in advance the elastic constants, visco elastic
constants, density as the coefficients of the prescribed first
order partial differential equations.
20. The elasticity and visco-elasticity constants measurement
apparatus of claim 14, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
21. The elasticity and visco-elasticity constants measurement
apparatus of claim 15, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
22. The elasticity and visco-elasticity constants measurement
apparatus of claim 16, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
23. The elasticity and visco-elasticity constants measurement
apparatus of claim 17, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
24. The elasticity and visco-elasticity constants measurement
apparatus of claim 18, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
25. The elasticity and visco-elasticity constants measurement
apparatus of claim 19, in which means of calculating elastic and
visco elastic constants utilizes at least one of measured in
advance or set in advance the reference elastic constants,
reference visco elastic constants, reference density as the initial
conditions of the prescribed first order partial differential
equations.
26. The elasticity and visco-elasticity constants measurement
apparatus of claim 14, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
27. The elasticity and visco-elasticity constants measurement
apparatus of claim 15, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
28. The elasticity and visco-elasticity constants measurement
apparatus of claim 16, in which means of calculating elastic-and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
29. The elasticity and visco-elasticity constants measurement
apparatus of claim 17, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
30. The elasticity and visco-elasticity constants measurement
apparatus of claim 18, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
31. The elasticity and visco-elasticity constants measurement
apparatus of claim 19, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
32. The elasticity and visco-elasticity constants measurement
apparatus of claim 20, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
33. The elasticity and visco-elasticity constants measurement
apparatus of claim 21, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
34. The elasticity and visco-elasticity constants measurement
apparatus of claim 22, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
35. The elasticity and visco-elasticity constants measurement
apparatus of claim 23, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
36. The elasticity and visco-elasticity constants measurement
apparatus of claim 24, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
37. The elasticity and visco-elasticity constants measurement
apparatus of claim 25, in which means of calculating elastic and
visco elastic constants utilizes regularized algebraic equations
together with finite difference method or finite element method as
numerical solution of the prescribed first order partial
differential equations.
38. The elasticity and visco-elasticity constants measurement
apparatus-based treatment apparatus equipped with the following
means: treatment transducer arrayed with more than one oscillator,
means (circuit) of treatment transmitting (transmitter of driving
signals to each oscillator of the treatment transducer array),
diagnosis transducer arrayed with more than one oscillator, means
(circuit) of diagnosis transmitting (transmitter of driving signals
to each oscillator of the diagnosis transducer array), means
(circuit) of receiving (receiver of the echo signals detected at
the oscillators of the transducers and matcher of the echo signals
based on their phases), means of calculating elastic and visco
elastic constants (calculator of at least one of elastic constants,
visco elastic constants, or density from the matched echo signals),
means of output of degeneration information on parts including the
lesions (output means of degeneration information based on
calculated at least one of the elastic constants, visco elastic
constants, or density), controller of the means (circuit) of
treatment transmitting, means (circuit) of diagnosis transmitting,
means (circuit) of receiving, and means of calculating elastic and
visco elastic constants, and the input means of commands and
conditions into the controller, wherein; the controller is not only
equipped with functions for controlling the means (circuit) of
diagnosis transmitting and means (circuit) of receiving based on
the commands and the conditions, but also with functions for
deforming the ROI in the target based on the commands and the
conditions, and for controlling the means (circuit) of treatment
transmitting to control the treatment ultrasound beam transmitted
from the treatment transducer based on the commands and the
conditions, and the means of calculating elastic and visco elastic
constants obtains the matched echo signals in the ROI based on the
commands given from the controller, and calculates at least one of
strain tensor data, strain rate tensor data, or acceleration vector
data in the ROI, and subsequently calculates from one of these
deformation data at least one of elastic constants, visco elastic
constants, or density in the ROI.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of the Invention
[0002] The present invention relates to an apparatus and method for
low-destructively (low-invasively) measuring mechanical properties
within object such as structures, substances, materials, living
tissues (liver, prostate, breast, etc.). For instance, measured can
be, due to applied stress and/or vibration by arbitrary mechanical
sources, generated displacement vector, strain tensor, strain rate
tensor, acceleration vector, or velocity vector, etc. within the
body. Furthermore, from these measured deformation data, following
constants can be measured, elastic constants such as shear modulus,
Poisson's ratio, etc., visoc elastic constants such as visco shear
modulus, visco Poisson's ratio, etc., delay times or relaxation
times relating these elastic constants and visco elastic constants,
or density.
[0003] On typical applied field, i.e., medical field, such as
ultrasonic diagnosis, nuclear magnetic resonance diagnosis, light
diagnosis, radio therapeutics, the present method and apparatus can
be applied for monitoring tissue degeneration, i.e., treatment
effectiveness. Otherwise, on structures, substances, materials,
living tissues, measured static and/or dynamic mechanical
properties can be utilized for evaluation, examination, diagnosis,
etc.
[0004] 2. Description of a Related Art
[0005] For instance, on medical field (liver, prostate, breast,
etc.), lesions are proposed to be treated by cryotherapy, or by
applying radioactive ray, high intensity focus ultrasound, laser,
electromagnetic RF wave, microwave, etc. In these cases, the
treatment effectiveness is proposed to be monitored. Moreover,
chemotherapy (anti-cancer drag etc.) effectiveness is also proposed
to be monitored. For instance, on radiotherapy etc., the treatment
effectiveness can be confirmed by low-invasively measuring
degeneration (including temperature change) of the lesion.
Otherwise, due to applied stress to the tissue part of interest
including lesions, generated deformations and deformation changes
are measured, from which the pathological state of the tissue is
evaluated such as elastic constants etc. Thus, based on the
measured distinct pathological state, the part of interest is
diagnosed, or treatment effectiveness is observed.
[0006] Temperature is known to have high correlations with elastic
constants, visco elastic constants, delay times or relaxation times
relating elastic constants and visco elastic constants, and
density, etc. Therefore, by measuring the following constants,
temperature distribution can be measured, i.e., elastic constants
such as shear modulus, Poisson's ratio, etc., visco elastic
constants such as visco shear modulus, visco Poisson's ratio, etc.,
delay times or relaxation times relating these elastic constants
and visco elastic constants, or density.
[0007] In the past, elastic constants and visco elastic constants
have been measured by applying stress at many points and by
measuring the responses such as stresses and strains. That is,
stress meter and strain meter are used, and sensitivity analysis is
numerically performed with utilization of finite difference method
or finite element method. Otherwise, in addition to elastic
constants, visco elastic constants such as visco shear modulus,
visco Poisson's ratio, etc. has been also measured by estimating
the shear wave propagation velocity generated by applying
vibrations.
[0008] As other monitoring techniques, temperature distribution is
measured by evaluating nuclear magnetic resonance frequency,
electric impedance, ultrasound velocity, etc. However, to measure
temperature, these techniques need other relating physical
properties of the target tissue. If degeneration occurs on the part
of region, the relating physical properties also change; thus
resulting severe limitations of the temperature measurement.
[0009] Other disadvantage is that the past measurement technique
needs many independent deformation fields generated by mechanical
sources outside the target body. However, if there exist internal
mechanical sources or mechanical sources are uncontrollable, the
technique becomes unavailable. That is, the past technique needs
all information about mechanical sources, such as positions, force
directions, force magnitudes, etc. Moreover, the technique needs
stress data and strain data at the target body surface, and needs
whole body model (finite difference method or finite element
method). Furthermore, low are spatial resolutions of measured
elastic constants and visco elastic constants from shear wave
velocity.
[0010] On the other hand, medical ultrasound diagnosis apparatus
can low-invasively image tissue distribution by converting
ultrasonic echo signals (echo signals) to image, after transmitting
ultrasonic pulses to target tissue and receiving the echo signals
at ultrasound transducer. Thus, by ultrasonically measuring tissue
displacements generated due to arbitrary mechanical sources, or by
measuring generated tissue strains, tissue elastic constants, etc.,
differences of between lesion and normal tissue can be observed
low-invasively. For instance, measured within the body can be, due
to applied stress and/or vibration by arbitrary mechanical sources,
generated displacement vector, strain tensor, strain rate tensor,
acceleration vector, velocity vector, etc. Furthermore, from these
measured deformation data, following constants can be measured,
elastic constants such as shear modulus, Poisson's ratio, etc.,
visco elastic constants such as visco shear modulus, visco
Poisson's ratio, etc., delay times or relaxation times relating
these elastic constants and visco elastic constants, or
density.
[0011] Then, in the past, tissue displacement has been proposed to
be measured to low-invasively diagnose tissue and lesion by
evaluating the echo signal changes of more than one time
transmitting signal. From the measured displacement distribution,
strain distribution is obtained, by which distribution of
pathological state of tissue has been proposed to be diagnosed
(Japanese Patent Application Publication JP-A-7-55775,
JP-A-2001-518342). Specifically, 3 dimensional (3D), 2D, or 1D
region of interest (ROI) are set in the target body, and
distributions of three, two, or one displacement component are
measured, from which in addition to strain tensor distribution,
elastic constant distributions, etc. are also numerically
obtained.
[0012] In addition to ultrasound transducer, as the displacement
(strain) sensor, utilized can be known contact or non-contact
sensors such as electromagnetic wave (including light) detector
etc. As mechanical sources, compressor and vibrator can be,
transducer-mounted apparatuses, not transducer-mounted ones,
internal heart motion, respiratory motion. If ROI is deformed by
ultrasound transmitted from sensor, there may not need other
mechanical sources except for the sensor. In addition to the
stationary elastic constants, difference of the tissue pathological
state includes dynamic changes of elastic constants and temperature
due to treatment.
[0013] However, as the classical tissue displacement measurement
methods assume tissue being deformed only in the axial (beam)
direction, when tissue also moves in lateral (scan) direction, the
classical method has low axial measurement accuracy. That is, the
displacement was determined only by 1D axial processing of
ultrasound echo signals.
[0014] Recently, displacement accuracy is improved by us through
development of 2D displacement vector measurement method, i.e.,
phase gradient estimation method of the 2D echo cross-spectrum
based on so-called the 2D cross-correlation processing and the
least squares processing. This method can suitably cope with
internal, uncontrollable mechanical sources (e.g., heart motion,
respiratory motion, blood vessel motion, body motion, etc.).
[0015] However, strictly speaking, measurement accuracy of actual
3D tissue displacement becomes low because the method can measure
by 2D processing of echo signals two displacement components or by
1D processing one displacement component.
[0016] Particularly, as lateral component of echo signal has a
narrow bandwidth and has no carrier frequency, lateral displacement
measurement accuracy and spatial resolution are much lower compared
with axial ones. Thus, the low lateral measurement accuracy
degrades the 3D displacement vector measurement and the 3D strain
tensor measurement.
[0017] Furthermore, when large displacement needs to be handled,
before estimating the gradient of the cross-spectrum phase, the
phase must be unwrapped, or the displacement must be coarsely
estimated by cross-correlation method as multiples of sampling
intervals. Thus, measurement process had become complex one.
SUMMARY OF THE INVENTION
[0018] The first purpose of the present invention is to provide an
apparatus and method for low-destructively measuring mechanical
properties within object such as structures, substances, materials,
living tissues (liver, prostate, breast, etc.), even if there
exists internal, or uncontrollable mechanical sources. The first
purpose of the present invention is, for instance, for diagnosing
and monitoring treatment effectiveness on living tissue, to provide
the measurement technique of following constants, elastic constants
such as shear modulus, Poisson's ratio, etc., visco elastic
constants such as visco shear modulus, visco Poisson's ratio, etc.,
delay times or relaxation times relating these elastic constants
and visco elastic constants, or density.
[0019] The second purpose of the present invention is to provide
the low-invasive treatment technique with utilization of
low-invasive measurement of the following constants, elastic
constants such as shear modulus, Poisson's ratio, etc., visco
elastic constants such as visco shear modulus, visco Poisson's
ratio, etc., delay times or relaxation times relating these elastic
constants and visco elastic constants, or density.
[0020] The third purpose of the present invention is to improve
measurement accuracy of displacement vector distribution generated
in 3D, 2D (including or not including beam direction), or 1D (beam
direction or scan direction) ROI in the target body when estimating
gradient of the echo cross-spectrum phase. Cross-spectrum can be
also estimated by Fourier's transform of echo cross-correlation
function.
[0021] The fourth purpose of the present invention is to simplify
calculation process into one without unwrapping the cross-spectrum
phase nor utilizing cross-correlation method; thus reducing
calculation amount and shortening calculation time.
[0022] The fifth purpose of the present invention is to improve
measurement accuracy of lateral displacements (orthogonal
directions to beam direction).
[0023] In the preferred embodiment of the present invention,
above-described purposes are achieved.
[0024] All the displacement measurement methods related to the
present invention allow measuring local displacement vector or
local displacement vector components from the phases of the
ultrasound echo signals acquired from the target as the responses
to more than one time transmitted ultrasound.
[0025] One method measures the displacement vector component from
the gradient of the cross-spectrum phase evaluated from echo
signals acquired at two different time, i.e., before and after
tissue deformation. The 3D processing yields from 3D cross-spectrum
phase .theta.(.omega.z,.omega.y,- .omega.z) accurate measurements
of 3D displacement vectors ((d=(dx, dy, dz).sup.T) in 3D ROI, and
consequently, results in measurements of the more accurate
displacement vector components compared with corresponding ones
measured by 2D processing (2D cross-spectrum phase:
.theta.(.omega.x,.omega.y), or .theta.(.omega.y,.omega.z), or
.theta.(.omega.x,.omega.z)) and 1D processing (1D cross-spectrum
phase: .theta.(.omega.x), or .theta.(.omega.y), or
.theta.(.omega.z)).
[0026] When measuring displacement from the gradient of the echo
cross-spectrum phase, to result the more accurate measurement
accuracy, the least squares method can be applied with utilization
as the weight function of the squares of the cross-spectrum usually
normalized by the cross-spectrum power, where, to stabilize the
measurement, the regularization method can be applied, by which a
priori information can be incorporated, i.e., about within the ROI
the magnitude of the unknown displacement vector, spatial
continuity and differentiability of the unknown displacement vector
distribution, etc. The regularization parameter depends on
time-space dimension of the ROI, direction of the unknown
displacement component, position of the unknown displacement
vector, etc. Otherwise, directional independent regularization
utilize the mechanical properties of tissue (e.g.,
incompressibility) and compatibility conditions of displacement
vector distribution and displacement component distribution.
[0027] The displacement measurement apparatus related to the
present invention can be equipped with the following means:
displacement (strain) sensor (transducer to transmit ultrasounds to
the target, and detect echo signals generated in the target),
relative position controller and relative direction controller
between the sensor and the target, means of transmitting/receiving
(transmitter of driving signals to the sensor, and receiver of the
echo signals detected at the sensor), means of data processing
(controller of the driving signals of the means of transmitting,
and processor of the received echo signals of means of receiving),
and means of data storing (storage of echo signals, measured
deformation data).
[0028] The means of data processing also measures the local
displacement vector or the local displacement vector components
utilizing the stated displacement measurement methods from the
phases of the ultrasound echo signals acquired from the target as
the responses to more than one time transmitted ultrasound.
[0029] The strain tensor measurement apparatus related to the first
point of view of the present invention can be equipped with the
displacement measurement apparatus, and the means of data
processing can yield strain tensor components by spatial
differential filtering with suitable cut off frequency in spatial
domain or frequency domain the measured 3D, or 2D displacement
vector components, or measured one direction displacement component
in the 3D, 2D, or 1D ROI. The means of data processing can also
yield strain rate tensor components, acceleration vector
components, or velocity vector components by time differential
filtering with suitable cut off frequency in time domain or
frequency domain the measured time series of displacement
components, or strain components.
[0030] The strain tensor measurement method related to the present
invention also allow directly measuring the local strain tensor or
the local strain tensor components from the phases of the
ultrasound echo signals acquired from the target as the responses
to more than one time transmitted ultrasound.
[0031] The strain tensor measurement apparatus related to the
second point of view of the present invention can be equipped with
the following means: displacement (strain) sensor (transducer to
transmit ultrasounds to the target, and detect echo signals
generated in the target), relative position controller and relative
direction controller between the sensor and the target, means of
transmitting/receiving (transmitter of driving signals to the
sensor, and receiver of the echo signals detected at the sensor),
means of data processing (controller of the driving signals of the
means of transmitting, and processor of the received echo signals
of means of receiving), and means of data storing (storage of echo
signals, measured deformation data).
[0032] The means of data processing also directly measures the
local strain tensor or the local strain tensor components utilizing
the stated direct strain measurement methods from the phases of the
ultrasound echo signals acquired from the target as the responses
to more than one time transmitted ultrasound.
[0033] The elasticity and visco-elasticity constants measurement
apparatus related to the first point of view of the present
invention can be equipped with the following means: means of data
storing (storage of at least one of strain tensor data, strain rate
tensor data, acceleration vector data, elastic constants, visco
elastic constants, or density measured in the ROI set in the
target), and means of calculating elastic and visco elastic
constants (calculator of at least one of elastic constants, visco
elastic constants, or density of arbitrary point in the ROI from at
least one of the measured strain tensor data, strain rate tensor
data, or acceleration vector data).
[0034] The means of calculating elastic and visco elastic constants
numerically determines at least one of the elastic constants, visco
elastic constants, or density from the first order partial
differential equations relating at least one of the elastic
constants, visco elastic constants, or density to at least one of
the strain tensor data, strain rate tensor data, acceleration
vector data. Time delays or relaxation times can also be determined
by ratio of the corresponding elastic constant and visco elastic
constant.
[0035] The elasticity and visco-elasticity constants measurement
apparatus related to the second point of view of the present
invention can be equipped with the following means: means of data
storing (storage of at least one of strain tensor data, strain rate
tensor data, acceleration vector data, elastic constants, visco
elastic constants, or density measured in the ROI including
lesions), means of calculating elastic and visco elastic constants
(calculator of at least one of elastic constants, visco elastic
constants, or density of arbitrary point in the ROI from at least
one of the measured strain tensor data, strain rate tensor data, or
acceleration vector data), and means of output of degeneration
information on parts including the lesions (output means of
degeneration information based on calculated at least one of the
elastic constants, visco elastic constants, or density).
[0036] The means of calculating elastic and visco elastic constants
numerically determines at least one of the elastic constants, visco
elastic constants, or the density from the first order partial
differential equations relating at least one of the elastic
constants, visco elastic constants, or density to at least one of
the strain tensor data, strain rate tensor data, acceleration
vector data.
[0037] The elasticity and visco-elasticity constants measurement
apparatus-based treatment apparatus related to the present
invention can be equipped w with the following means: treatment
transducer arrayed with more than one oscillator, means (circuit)
of treatment transmitting (transmitter of driving signals to each
oscillator of the treatment transducer array), diagnosis transducer
arrayed with more than one oscillator, means (circuit) of diagnosis
transmitting (transmitter of driving signals to each oscillator of
the diagnosis transducer array), means (circuit) of receiving
(receiver of the echo signals detected at the oscillators of the
transducers and matcher of the echo signals based on their phases),
means of calculating elastic and visco elastic constants
(calculator of at least one of elastic constants, visco elastic
constants, or density from the matched echo signals), means of
output of degeneration information on parts including the lesions
(output means of degeneration information based on calculated at
least one of the elastic constants, visco elastic constants, or
density), controller of the means (circuit) of treatment
transmitting, means (circuit) of diagnosis transmitting, means
(circuit) of receiving, and means of calculating elastic and visco
elastic constants, and the input means of commands and conditions
into the controller.
[0038] The controller is not only equipped with functions for
controlling the means (circuit) of diagnosis transmitting and means
(circuit) of receiving based on the commands and the conditions,
but also with functions for deforming the ROI in the target based
on the commands and the conditions, and for controlling the means
(circuit) of treatment transmitting to control the treatment
ultrasound beam transmitted from the treatment transducer based on
the commands and the conditions.
[0039] The means of calculating elastic and visco elastic constants
obtains the matched echo signals in the ROI based on the commands
given from the controller, and calculates at least one of strain
tensor data, strain rate tensor data, or acceleration vector data
in the ROI, and subsequently calculates from these deformation data
at least one of elastic constants, visco elastic constants, or
density in the ROI. Here, controlled of treatment ultrasound beam
may be beam focus position, treatment interval, ultrasound beam
power, ultrasound beam strength, transmit term, beam shape
(apodization), etc. The oscillators may serve both as treatment
ones and diagnosis ones.
BRIEF DESCRIPTION OF THE DRAWING
[0040] FIG. 1 shows a schematic representation of a global frame of
displacement vector and strain tensor measurement apparatus, and
elasticity and visco-elasticity constants measurement apparatus,
related to one of conduct forms of the present invention;
[0041] FIG. 2 shows illustration of a displacement (strain) sensor
applicable to the present invention;
[0042] FIG. 3 shows illustration of mechanical scan movements of
the displacement (strain) sensor;
[0043] FIG. 4 shows illustration of beam steering, and spatial
interporation of measured two displacement vector component
distributions;
[0044] FIG. 5 shows illustration of transmitted beams whose
amplitudes are sine modulated in scan directions;
[0045] FIG. 6 shows illustration of a basic (n=) wave component and
n-th harmonic wave components (n equals from 2 to N) of ultrasound
echo signal;
[0046] FIG. 7 shows illustration of a local 3D space centered on a
point (x,y,z) in 3D ROI in pre-deformation ultrasound echo signal
space, and the shifted one in post-deformation ultrasound echo
signal space;
[0047] FIG. 8 shows illustration as the example of searching for
local 3D ultrasound echo signal by phase matching in searching
space set in post-deformation ultrasound echo signal space. That
is, the corresponding local signal is searched for using
pre-deformation local echo signal;
[0048] FIG. 9 shows illustration to make 3D displacement vector
distribution high spatial resolution, i.e., to make local space
small;
[0049] FIG. 10 shows flowchart of method of 3D displacement vector
distribution in 3D space (method 1-1), of method of 2D displacement
vector distribution in 2D region (method 2-1), of method of one
direction displacement component distribution in 1D region (method
3-1);
[0050] FIG. 11 shows flowchart of method of 3D displacement vector
distribution in 3D space (method 1-2), of method of 2D displacement
vector distribution in 2D region (method 2-2), of method of one
direction displacement component distribution in 1D region (method
3-2);
[0051] FIG. 12 shows flowchart of method of 3D displacement vector
distribution in 3D space (method 1-3), of method of 2D displacement
vector distribution in 2D region (method 2-3), of method of one
direction displacement component distribution in 1D region (method
3-3);
[0052] FIG. 13 shows flowchart of method of 3D displacement vector
distribution in 3D space (method 1-4), of method of 2D displacement
vector distribution in 2D region (method 2-4), of method of one
direction displacement component distribution in 1D region (method
3-4);
[0053] FIG. 14 shows flowchart of method of 3D displacement vector
distribution in 3D space (method 1-5), of method of 2D displacement
vector distribution in 2D region (method 2-5), of method of one
direction displacement component distribution in 1D region (method
3-5);
[0054] FIG. 15 shows illustration of a local 2D region centered on
a point (x,y) in 2D ROI in pre-deformation ultrasound echo signal
space, and the shifted one in post-deformation ultrasound echo
signal space;
[0055] FIG. 16 shows illustration as the example of searching for
local 2D ultrasound echo signal by phase matching in searching
region set in post-deformation ultrasound echo signal space. That
is, the corresponding local signal is searched for using
pre-deformation local echo signal;
[0056] FIG. 17 shows illustration to make 2D displacement vector
distribution high spatial resolution, i.e., to make local region
small;
[0057] FIG. 18 shows illustration of a local 1D region centered on
a point (x) in 1D ROI in pre-deformation ultrasound echo signal
space, and the shifted one in post-deformation ultrasound echo
signal space;
[0058] FIG. 19 shows illustration as the example of searching for
local 1D ultrasound echo signal by phase matching in searching
region set in post-deformation ultrasound echo signal space. That
is, the corresponding local signal is searched for using
pre-deformation local echo signal;
[0059] FIG. 20 shows illustration to make one direction
displacement component distribution high spatial resolution, i.e.,
to make local region small;
[0060] FIG. 21 shows flowchart of method of 2D displacement vector
distribution in 3D space (method 4-1), of method of one direction
displacement component distribution in 3D space (method 5-1), and
of method of one direction displacement component distribution in
2D region (method 6-1);
[0061] FIG. 22 shows flowchart of method of 2D displacement vector
distribution in 3D space (method 4-2), of method of one direction
displacement component distribution in 3D space (method 5-2), and
of method of one direction displacement component distribution in
2D region (method 6-2);
[0062] FIG. 23 shows flowchart of method of 2D displacement vector
distribution in 3D space (method 4-3), of method of one direction
displacement component distribution in 3D space (method 5-3), and
of method of one direction displacement component distribution in
2D region (method 6-3);
[0063] FIG. 24 shows flowchart of method of 2D displacement vector
distribution in 3D space (method 4-4), of method of one direction
displacement component distribution in 3D space (method 5-4), and
of method of one direction displacement component distribution in
2D region (method 6-4);
[0064] FIG. 25 shows flowchart of method of 2D displacement vector
distribution in 3D space (method 4-5), of method of one direction
displacement component distribution in 3D space (method 5-5), and
of method of one direction displacement component distribution in
2D region (method 6-5);
[0065] FIG. 26 shows flowchart of measurement procedure of
elasticity constants, and visco-elasticity constants utilizing the
elasticity and visco-elasticity constants measurement apparatus
(FIG. 1);
[0066] FIG. 27 shows a schematic representation of a global frame
of elasticity and visco-elasticity constants measurement
apparatus-based treatment apparatus related to one of conduct forms
of the present invention; and
[0067] FIG. 28 shows flowchart of control procedure of the
elasticity and visco-elasticity constants measurement
apparatus-based treatment apparatus (FIG. 27).
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0068] The following is explanation in detail of conduct forms of
the present invention with referring to figures.
[0069] FIG. 1 shows a schematic representation of a global frame of
displacement vector and strain tensor measurement apparatus, and
elasticity and visco-elasticity constants measurement apparatus,
related to one of conduct forms of the present invention. This
apparatus measures in 3D, 2D, or 1D ROI 7 set in measurement object
6 displacement vector component distributions, strain tensor
component distributions, their time-space partial derivative
distributions, etc. to obtain strain tensor field, strain rate
tensor field, acceleration vector etc., from which this apparatus
measures following constant distributions, i.e., elastic constants
such as shear modulus, Poisson's ratio, etc., visco elastic
constants such as visco shear modulus, visco Poisson's ratio, etc.,
delay times or relaxation times relating these elastic constants
and visco elastic constants, or density.
[0070] As shown in FIG. 1, displacement (strain) sensor 5 can be
directly contacted to object surface, or suitable medium can be put
between the sensor and the object. On this conduct form, as the
displacement (strain) sensor, ultrasound transducer is used. The
transducer may have 1D or 2D array of oscillators.
[0071] Distance between the object 6 and the displacement (strain)
sensor 5 can be mechanically controlled by position controller 4.
Moreover, relative distance between the object 6 and the
displacement (strain) sensor 5 can be mechanically controlled by
position controller 4. Ultrasound transmitter (ultrasound pulser)
5, is equipped to drive the displacement (strain) sensor 5, and 5'
also serves as output controller, i.e., receiver with amplifier of
echo signals detected at the displacement (strain) sensor 5.
Furthermore, mechanical source 8 can be equipped to actively apply
static compression, vibration, etc., and mechanical position
controller 4" can be also equipped.
[0072] Output echo signals of output controller 5' are stored at
storage 2 passing through measurement controller 3. The echo
signals stored at storage 2 are read out by data processor 1, and
displacement vector component distributions (time series) or strain
tensor component distributions (time series) are directly
calculated and obtained of arbitrary time in the ROI 7, and further
calculated and obtained are as their time-space partial
derivatives, i.e., strain tensor component distributions (time
series), strain rate tensor component distributions (time series),
acceleration vector component distributions (time series), etc.
That is, when displacement vector component distributions are
calculated of the ROI 7, strain tensor component distributions
(time series) are obtained by implementing 3D, 2D, or 1D spatial
differential filter to the obtained displacement vector component
distributions (time series). The cut off frequencies of all the
filters used in the present invention can be set different values
freely at each point at each time in each spatio-temporal direction
as those of usual filters. The acceleration vector component
distributions (time series) are obtained by implementing time
differential filter twice to the measured displacement vector
component distributions (time series). The strain rate tensor
component distributions (time series) are obtained-by implementing
spatial differential filter to the velocity vector component
distributions (time series) obtained by implementing time
differential filter to the displacement vector component
distributions (time series), or by implementing time differential
filter once to the measured strain tensor component distributions
(time series). Moreover, when strain tensor component distributions
(time series) are directly calculated of the ROI 7 and obtained,
strain rate tensor component distributions (times series) are
obtained by implementing time differential filter to the measured
strain tensor component distributions (time series). Furthermore,
this data processor 1 calculates following constant distributions,
i.e., elastic constants such as shear modulus, Poisson's ratio,
etc., visco elastic constants such as visco shear modulus, visco
Poisson's ratio, etc., delay times or relaxation times relating
these elastic constants and visco elastic constants, or density
from the measured distributions of strain tensor components (time
series), strain rate tensor components (time series), acceleration
vector components (time series), etc. These calculated results are
stored at the storage 2.
[0073] The measurement controller 3 controls the data processor 1,
the position controller 4 and 4", and the transmitting/output
controller 5'. The position controller 4' is not utilized when the
object 6 is spatially fixed. When displacement (strain) sensor 5 is
electronic scan type, position controller 4 is not always utilized.
That is, it may be possible to measure without mechanical scanning.
The displacement (strain) sensor 5 may be contacted on the object
6, or may not. That is, the displacement (strain) sensor 5 and the
object 6 may be dipped in or immersed in water tank, for instance,
when monitoring the treatment effectiveness of High Intensity Focus
Ultrasound (HIFU).
[0074] The position controller 4 mechanically controls the relative
position between the displacement (strain) sensor and the object 6.
Specifically, the position controller 4 realizes vertical, horizon,
turn, and fan direction scan movements (FIG. 3). The output of the
transmitting/output controller 5' is also stored at storage 2
successively or with given time intervals. The data processor 1
controls the transmitting/output controller 5', and acquires the
echo's basic wave components, n-th harmonic wave components (n
equals from 2 to N), or all the components in 3D, 2D, or 1D ROI 7,
and implements below-described data processing to yield
displacement data, strain data, strain rate data, or acceleration
data, and stores measured these data in the storage 2.
[0075] The transmitting/output controller 5' and the data processor
1 obeys the commands of measurement controller 3, and carry out
synthetic aperture processing, e.g., transmitting fixed focusing
process, multi-transmitting fixed focusing process, receiving
dynamic focusing process, etc. Furthermore, the transmitting/output
controller 5' and the data processor 1 carry out apodization
process of ultrasound signals, i.e., weighting process on each
ultrasound transmitted/received at each oscillator to sharpen the
synthesized ultrasound beam, and carry out beam steering process to
acquire the echo signals of 3D, 2D, or 1D ROI.
[0076] Next explanation is in detail about displacement and strain
measurement apparatus related to conduct forms of the present
invention.
[0077] On this conduct form, as the displacement (strain) sensor 5,
the following type ultrasound transducers can be utilized, i.e., 2D
array being mechanical scan possible, 2D array being electronic
scan possible, 1D array being mechanical scan possible, and 1D
array being electronic scan possible.
[0078] On this conduct form, synthetic aperture can be performed.
Also beam steering can be performed. When beam steering is
performed, measured displacement component distributions and strain
tensor component distributions are spatially interporated, after
which these measured displacement component distributions (time
series) and strain tensor component distributions (time series) are
time-spatially differentiated to yield strain tensor component
distributions (time series), strain rate tensor component
distributions (time series), acceleration vector component
distributions (time series), and velocity vector component
distributions (time series).
[0079] As measurement of the beam direction is considerably
accurate compared with that of the orthogonal scan direction, to
yield high accuracy displacement vector measurement, mechanical
scan and/or beam steering can be performed. That is, echo data
frames are acquired by performing mechanical scan and/or beam
steering such that ultrasound beams are transmitted in three
different directions when measuring 3D displacement vector, and in
two different directions when measuring 2D displacement vector.
From two echo data frames acquired by transmitting the ultrasound
beams in same direction, accurately the distribution of
displacement component in beam direction is measured, by which
accurate 3D or 2D displacement vector distribution can be obtained
(e.g., FIG. 4).
[0080] However, to obtain the final displacement vector
distribution, displacement vector distributions on the different
old discrete coordinates must be converted to ones on one new
discrete coordinate. That is, by interporating the displacement
component distribution measured on the old discrete coordinate, the
displacement component can be obtained at the point of the new
discrete coordinate. Concretely, displacement component
distribution is Fourier's transformed, which is multiplied with
complex exponential such that the phase is shifted. Thus, realized
is spatial shifting of the displacement component distribution.
[0081] On this conduct form, amplitudes of transmitted beams can be
sine modulated in scan directions.
[0082] The sine modulation frequency is better to be higher.
However, as this modulation shifts in frequency domain in scan
direction the band determined by beam width, based on the sampling
theorem the modulation frequency needs to be set such that the
highest frequency becomes less than the half of the sampling
frequency determined by beam pitch. Thus, improved is measurement
accuracy of displacement component distribution in scan direction
being orthogonal to beam direction.
[0083] Based on these processes, obtained ultrasound echo signals
in 3D, 2D, or 1D ROI can be effectively utilized, i.e., basic wave
components, harmonic wave components (The carrier frequency higher,
improved is measurement accuracy of displacement component in beam
direction. The carrier frequency higher, the beam width narrower.
Thus, as the bandwidth in scan direction is wider compared with the
basic component wave, also improved is measurement accuracy of
displacement component in scan direction.), or all the wave
components due to low SNRs of only harmonic wave components.
[0084] That is, below-described displacement and strain measurement
methods can utilize the ultrasound echo signals, or only extracted
the basic wave components, or only extracted the n-th harmonic wave
components (n equals from 2 to N), or these combinations (methods
from 1-1 to 1-5, from 2-1 to 2-5, from 3-1 to 3-5, from 4-1 to 4-5,
from 5-1 to 5-5, from 6-1 to 6-5.).
[0085] These stated displacement and strain measurement methods
base on iteratively updating the displacement estimate utilizing
the estimated remaining error data (estimated residual displacement
data). The initial estimate is set based on the a priori knowledge
about measurement target, i.e., displacement distribution, strain
distribution, strain rate distribution, acceleration distribution,
or velocity distribution. Finally obtained are accurate
displacement vector distribution (time series), displacement vector
component distributions (time series), strain tensor distribution
(time series), strain tensor component distributions (time series),
strain rate tensor distribution (time series), strain rate tensor
component distributions (time series), acceleration vector
distribution (time series), acceleration vector component
distributions (time series) velocity vector distribution (time
series), or velocity vector component distributions (time
series).
[0086] However, when stressing on real-time processing, measurement
can be finished only with the once estimation.
[0087] During iterative estimation of the displacement vector and
residual displacement vector, when estimation errors are detected a
priori as the points of time-space magnitude and time-space
continuity, for instance, the estimates can be cut by compulsion
such that the estimates range from the given smallest value to the
given largest value, or such that the difference between the
estimates of the neighboring points settle within the given
ranges.
[0088] On these stated iterative displacement and strain
measurement methods, all the methods for estimating the residual
displacement vector component or the displacement vector component
utilize as the index the phases of the ultrasound echo signals
acquired at more than one time. First of all, one of these methods
is used to explain the iterative methods, i.e., the method
estimating displacement from the gradient of the phase of the
cross-spectrum of ultrasound echo signals acquired twice.
[0089] The displacement and strain measurement methods can be
implemented each on extracted the basic wave signals and the n-th
harmonic wave components (n equals from 2 to N). In this case, the
final measurement result can be obtained as the mean displacement
data weighted by the power ratio of the cross-spectrums etc.
[0090] In addition, when measuring the displacement from the
gradient of the cross-spectrum phase utilizing least squares
method, data processor also utilize the regularization method based
on the a priori knowledge, which improves stability, accuracy, and
spatial resolutions of the measurement of the displacement vector
distribution, or the displacement vector component
distributions.
[0091] In the past, when large displacement needs to be handled,
before estimating the gradient of the cross-spectrum phase, the
phase had been unwrapped, or the displacement had been coarsely
estimated by cross-correlation method. Thus, measurement procedure
had become complex one. To cope with these complexity, the
measurement procedure is made simpler without these processes by
introducing process of thinning out data and process of remaking
data interval original. Thus, implemented soft amount and
calculation time are reduced. Occasionally, the regularization may
not be performed.
[0092] However, as other method, before estimating the gradient of
the cross-spectrum phase, the phase can also be unwrapped, or the
displacement can also be coarsely estimated by cross-correlation
method. Also in this case, when measuring the local displacement
from the gradient of the cross-spectrum phase, a priori knowledge
about the displacement distribution in the ROI can be incorporated
by utilizing the regularization method, where the least squares
method utilizes as the weight function the squares of the
cross-spectrum usually normalized by the cross-spectrum power.
Freely, when estimating the gradient of the cross-spectrum phase,
acquired ultrasound echo signals can be thinned out in each
direction with constant intervals.
[0093] These cases handles the gradient of the local 3D, 2D or 1D
cross-spectrum phase evaluated on 3D, 2D, or 1D ultrasound echo
signals acquired at more than one time from 3D space, 2D or 1D
region in the object. Stably measured with high accuracy and high
spatial resolutions are 3D displacement vector component
distributions in the 3D SOI (space of interest), 2D displacement
vector component distributions in the 2D ROI, one direction
displacement component distribution in the 1D ROI, 2D displacement
vector component distributions or one direction displacement
component distribution in the 3D SOI, or one direction displacement
component distribution in the 2D ROI.
[0094] The displacement and strain measurement apparatus of the
present invention measures in the 3D SOI, 2D, or 1D ROI in the
object the displacement vector distribution, the strain tensor
distribution, the strain rate tensor distribution, the acceleration
vector distribution, velocity vector distribution, etc. from
ultrasound echo signals measured in 3D SOI, 2D, or 1D ROI (referred
to 3D, 2D, 1D ultrasound echo signals). The displacement and strain
measurement apparatus can be equipped with:
[0095] displacement (strain) sensor (ultrasound transducer)
[0096] relative position controller and relative direction
controller between the sensor and the target (vertical, horizon,
turn, and fan direction scan movements), transmitter (ultrasound
pulser)/output controller (receiver and amplifier),
[0097] means of data processing (synthetic aperture process:
transmitting fixed focusing process, multi-transmitting fixed
focusing process, receiving dynamic focusing process etc.,
apodization),
[0098] means of data storing (storage of echo signals), means of
(signal) data processing (calculation of displacement vector
distribution, strain tensor distribution, strain rate tensor
distribution, acceleration vector distribution, velocity vector
distribution, etc.), and
[0099] means of data storing (storage of the displacement vector
distribution, strain tensor distribution, strain rate tensor
distribution, acceleration vector distribution, velocity vector
distribution, etc.).
[0100] In this case, the means of data processing can yield strain
tensor components by implementing spatial 3D, 2D, or 1D
differential filter with cut off frequency or multiplying Fourier's
transform of the differential filter in frequency domain to 3D
displacement vector component distributions in the 3D SOI (space of
interest), 2D displacement vector component distributions in the 2D
ROI, one direction displacement component distribution in the 1D
ROI, 2D displacement vector component distributions or one
direction displacement component distribution in the 3D SOI, or one
direction displacement component distribution in the 2D ROI.
Moreover, by implementing time differential filter with cut off
frequency or multiplying Fourier's transform of the differential
filter in frequency domain to time series of these, the strain rate
tensor component distributions, acceleration vector component
distributions, velocity vector component distributions. Moreover,
the strain rate tensor component distributions can be obtained from
directly measured strain tensor component distributions.
[0101] The displacement and strain measurement apparatus can be
also equipped with static compressor or vibrator as mechanical
source to generate at least one strain tensor field (one
displacement vector field) in the 3D SOI, 2D, or 1D ROI in the
object. On this case, generated due to body motion (heart motion,
blood vessel motion, respiratory), the strain tensor field
(displacement vector field) can be also measured in the 3D SOI, 2D,
or 1D ROI in the object.
[0102] The following ultrasound transducer type can be utilized,
i.e., ultrasound oscillator being mechanical scan possible,
electronic scan type 2D ultrasound oscillator array (being
mechanical scan possible), and 1D ultrasound oscillator array
(being mechanical scan possible). Thus, echo signal is synthesized
one. When the displacement (strain) sensor is contacted on the
object, the contact part can become mechanical source. That is, the
displacement (strain) sensor also serves as compressor or vibrator.
When the part of lesion is dipped in or immersed in water tank to
carry out treatment with High Intensity Focus Ultrasound (HIFU),
the object can be non-contactly measured by dipping in or immersing
the displacement (strain) sensor as well in water tank.
[0103] Moreover, when the displacement (strain) sensor is directly
contacted to object surface as mechanical source to stably measure
elastic constant distributions and visco elastic constant
distributions, suitable reference medium can be put between the
sensor and the object. In this case, the reference medium can also
be mounted (installed) on the transducer.
[0104] Basically, the means of data processing can yield strain
tensor component distributions, stain rate tensor component
distributions, acceleration vector component distributions, or
velocity vector component distributions from the obtained
deformation data utilizing the displacement (strain) sensor from
synthesized ultrasound echo in 3D SOI, 2D or 1D ROI, i.e., 3D
displacement vector component distributions in the 3D SOI, 2D
displacement vector component distributions in the 2D ROI, one
direction displacement component distribution in the 1D ROI, 2D
displacement vector component distributions or one direction
displacement component distribution in the 3D SOI, or one direction
displacement component distribution in the 2D ROI. Moreover, the
strain rate tensor component distributions can be obtained from
directly measured strain tensor component distributions.
[0105] In this case, the means of data processing can yield
displacement component distributions and strain tensor component
distributions from ultrasound echo signals acquired in each
dimensional ROI with beam steering as well as synthetic aperture
processing, from which obtained can be strain tensor component
distributions, strain rate tensor component distributions,
acceleration vector component distributions, and velocity vector
component distributions.
[0106] Moreover, in this case, the means of data processing can
yield displacement component distributions and strain tensor
component distributions from ultrasound echo basic wave components,
ultrasound echo harmonic wave components, or all the ultrasound
echo components, from which obtained can be strain tensor component
distributions, strain rate tensor component distributions,
acceleration vector component distributions, and velocity vector
component distributions.
[0107] Here, the sine modulation frequency is better to be higher.
However, as this modulation shifts in scan direction in frequency
domain the band determined by beam width, based on the sampling
theorem the modulation frequency needs to be set such that the
highest frequency becomes less than the half of the sampling
frequency determined by beam pitch.
[0108] Furthermore, ultrasound echo signals can be acquired by
combining the processing, i.e., synthetic aperture processing, beam
steering, sine modulation of transmitted beams' amplitudes in scan
directions. In this case, measured can be displacement vector
component distribution from ultrasound echo basic wave components,
ultrasound echo harmonic wave components, or all the ultrasound
echo components.
[0109] When utilizing below-described displacement and strain
measurement methods, as measurement of the beam direction is
considerably accurate compared with that of the orthogonal scan
direction, to yield high accuracy displacement measurements,
mechanical scan and/or beam steering are performed. That is, echo
data frames are acquired under object's pre- and post-deformation
by performing mechanical scan and/or beam steering such that
ultrasound beams are transmitted in three different directions when
measuring 3D displacement vector, and in two different directions
when measuring 2D displacement vector. From two echo data frames
acquired by transmitting the ultrasound beams in same direction,
accurately the distribution of displacement component in beam
direction is measured, by which accurate 3D or 2D displacement
vector distribution is obtained. To obtain the final displacement
vector distribution, displacement vector distributions on the
different old discrete coordinates must be converted to ones on one
new discrete coordinate. That is, by interporating the displacement
component distribution measured on the old discrete coordinate, the
displacement component can be obtained at the point of the new
discrete coordinate. Concretely, displacement component
distribution is Fourier's transformed, which is multiplied with
complex exponential such that the phase is shifted. Thus, realized
is spatial shifting of the displacement component distribution.
Strain tensor component distributions can be obtained from these
displacement measurement data. Moreover, from these time series,
obtained can be strain tensor rate component distributions,
acceleration vector component distributions, velocity vector
component distributions. Other displacement measurement methods and
strain measurement methods can be also applied to the ultrasound
echo time series data in similar ways.
[0110] Next explanation is in detail about displacement and strain
measurement algorithm related to conduct forms of the present
invention. The means of data processing 1 always carries out the
below-explained calculation process or their combination, or as
occasion demands.
[0111] (1) Calculation process of 3D displacement vector component
distribution in 3D ROI (below-described methods from 1-1 to
1-5)
[0112] (2) Calculation process of 2D displacement vector component
distribution in 2D ROI (below-described methods from 2-1 to
2-5)
[0113] (3) Calculation process of 1D (one direction) displacement
component distribution in 1D ROI (below-described methods from 3-1
to 3-5)
[0114] (4) Calculation process of 2D displacement vector component
distribution in 3D ROI (below-described methods from 4-1 to
4-5)
[0115] (5) Calculation process of 1D (one direction) displacement
component distribution in 3D ROI (below-described methods from 5-1
to 5-5)
[0116] (6) Calculation process of 1D (one direction) displacement
component distribution in 2D ROI (below-described methods from 6-1
to 6-5)
[0117] When beam steering is performed, at means data processing 1,
measured displacement vector component distributions are spatially
interporated.
[0118] With respect to displacement component distributions and
strain component distributions obtained through the above
calculation processes, means of data processing 1 performs
differentiation such that the followings are obtained, i.e., at
each time strain tensor component distributions, strain gradient
component distributions, strain rate tensor component
distributions, strain rate gradient component, acceleration vector
component distributions, or velocity vector distributions. These
calculated results are stored at storage 2. Moreover, these
calculated results are displayed on display apparatus such as CRT
(color or gray scaled) in real-time or in quasi real-time.
[0119] As static or motion image, or time course (difference)
image, etc., the followings can be displayed, i.e., displacement
vector distribution, displacement vector component distributions,
strain tensor component distributions, strain gradient component
distributions, strain rate tensor component distributions, strain
rate gradient component, acceleration vector component
distributions, or velocity vector component distributions. At
arbitrary points the values and their graph (time course) can also
be displayed. For instance, by utilizing ultrasound diagnosis
apparatus, spatial variations of bulk modulus and density of
tissues can be displayed in real-time. Thus, the above-described
static or motion image, or time course image of the displacement
vector distribution, etc. can also be superimposed and displayed on
the ultrasound image. The followings can be displayed in vector
style as well, i.e., the displacement vector distribution,
acceleration vector, velocity vector.
[0120] The following is explanation in detail of displacement
measurement and calculation processes.
[0121] (I) Method 1: Measurement of 3D Displacement Vector
Distribution
[0122] 3D displacement vector distribution can be measured in 3D
SOI 7 in the Cartesian coordinate system. 3D ultrasound echo
signals are acquired under pre-deformation and post-deformation.
These echo signals are processed by the below-described methods
1-1, 1-2, 1-3, 1-4, and 1-5. That is, as shown in FIG. 7, local
space is set at each point in the pre- and post-deformation 3D echo
signal, and as shown in FIG. 8, the corresponding local space is
iteratively searched for in the SOI 7 using the local phase
characteristics as the index. In this searching scheme, the
estimated residual displacement vector is used to update the
previously estimated displacement vector. When the estimated
residual displacement vector satisfies with prescribed condition,
the local space size is made small (FIG. 9). Thus, accurate 3D
displacement vector measurement is realized. Here, sampling
intervals are .DELTA.x, .DELTA.y, .DELTA.z respectively in the x,
y, and z-axes.
[0123] [Method 1-1]
[0124] The procedure of the method 1-1 is shown in FIG. 10. The
processes from 1 to 5 yields 3D displacement vector d(x,y,z)
[=(dx(x,y,z), dy(x,y,z), dz(x,y,z)).sup.T] of arbitrary point
(x,y,z) in 3D SOI from pre- and post-deformation local 3D echo
signals r.sub.1(l,m,n) and r.sub.2(l,m,n) [0.ltoreq.l.ltoreq.L-1,
0.ltoreq.m.ltoreq.M-1, 0.ltoreq.n.ltoreq.N-1] centered on (x,y,z)
of pre- and post-deformation 3D echo signals r.sub.1(x,y,z) and
r.sub.2(x,y,z). L, M, and N must be determined such that .DELTA.xL,
.DELTA.yM, .DELTA.zN are respectively at least 4 times longer than
corresponding displacement components
.vertline.dx(x,y,z).vertline., .vertline.dy(x,y,z).vertline.,
.vertline.dz(x,y,z).vertline..
[0125] (Process 1: Phase Matching at the Point (x,y,z))
[0126] Phase matching is performed to obtain i-th estimate
d.sup.i(x,y,z) [=(d.sup.ix(x,y,z), d.sup.iy(x,y,z),
d.sup.iz(x,y,z)).sup.T] of the 3D displacement vector d(x,y,z)
[=(dx(x,y,z), dy(x,y,z), dz(x,y,z)).sup.T].
[0127] Searching space is set in the post-deformation echo signal
space r.sub.2 (x,y,z), being centered on the local space
[0.ltoreq.1.ltoreq.L-1, 0.ltoreq.m.ltoreq.M-1,
0.ltoreq.n.ltoreq.N-1] centered on (x,y,z) and being twice longer
than the corresponding length, in order to update the i-1 th
estimate d.sup.i-1 (x,y,z) [=(d.sup.i-1x(x,y,z), d.sup.i-1y(x,y,z),
d.sup.i-1z(x,y,z)).sup.T] of the 3D displacement vector
d(x,y,z)[=(dx(x,y,z), dy(x,y,z), dz(x,y,z)).sup.T], where
d.sup.0(x,y,z)={haeck over (d)}(x,y,z). (1)
[0128] The phase of the post-deformation local echo signal is
matched to pre-deformation local echo signal by multiplying 1 exp {
j 2 L d x i - 1 ( x , y , z ) x 1 + j 2 M d y i - 1 ( x , y , z ) y
m + j 2 N d z i - 1 ( x , y , z ) z n } ( 2 )
[0129] to 3D Fourier's transform of this searching space echo
signal r'.sub.2(l,m,n) [0.ltoreq.l.ltoreq.2L-1,
0.ltoreq.m.ltoreq.2M-1, 0.ltoreq.n.ltoreq.2N-1] using i-th estimate
d.sup.i-1(x,y,z), or by multiplying 2 exp { j 2 L u ^ x i - 1 ( x ,
y , z ) x l + j 2 M u ^ y i - 1 ( x , y , z ) y m + j 2 N u ^ z i -
1 ( x , y , z ) z n } ( 2 ' )
[0130] to 3D Fourier's transform of the i-1 th phase-matched
searching space echo signal r'.sup.i-1.sub.2 (l,m,n) using the
estimate .sup.i-1 (x,y,z)[=(.sub.x.sup.i-1 (x,y,z), .sub.y.sup.i-1
(x,y,z), .sub.z.sup.i-1 (x,y,z)).sup.T] [.sup.0(x,y,z)=0 (zero
vector)]of the vector u.sup.i-1 (x,y,z) [=(u.sup.i-1.sub.x(x,y,z),
u.sup.i-1.sub.y(x,y,z), u.sup.i-1.sub.z (x,y,z)).sup.T].
[0131] By carrying out inverse Fourier's transform of this product,
post-deformation echo signal r.sup.i.sub.2 (l,m,n) is obtained at
the center of the searching space echo signal r'.sup.i.sub.2
(l,m,n) which is used at i-th stage to estimate 3D displacement
vector d(x,y,z)[=(dx(x,y,z), dy(x,y,z), dz(x,y,z)).sup.T].
[0132] Alternatively, the phase of the pre-deformation local echo
signal can be matched to post-deformation local echo signal in a
similar way. That is, 3D Fourier's transform of the searching space
echo signal r'.sub.1(l,m,n) [0.ltoreq.l.ltoreq.2L-1,
0.ltoreq.m.ltoreq.2M-1, 0.ltoreq.n.ltoreq.2N-1] centered on the
point (x,y,z) in the pre-deformation echo signal space is
multiplied with 3 exp { - j 2 L d x i - 1 ( x , y , z ) x l - j 2 M
d y i - 1 ( x , y , z ) y m - j 2 N d z i - 1 ( x , y , z ) z n } ,
( 2 " )
[0133] or 3D Fourier's transform of the i-1 th phase-matched
searching space echo signal r'.sup.i-1.sub.1 (l,m,n) is multiplied
with 4 exp { - j 2 L u ^ x i - 1 ( x , y , z ) x l - j 2 M u ^ y i
- 1 ( x , y , z ) y m - j 2 N u ^ z i - 1 ( x , y , z ) z n } . ( 2
"' )
[0134] (Process 2: Estimation of 3D Residual Displacement Vector at
the Point (x,y,z))
[0135] Local 3D echo cross-spectrum is evaluated from the 3D
Fourier's transforms of the pre-deformation local 3D ultrasound
echo signal r.sub.1 (l,m,n) and phase-matched post-deformation
local 3D ultrasound echo signal r.sup.i.sub.2 (l,m,n)
S.sup.i.sub.2,1(l,m,n)=R.sub.2.sup.i*(l,m,n)R.sub.1(l,m,n), (3)
[0136] where * denotes conjugate.
[0137] Alternatively, when pre-deformation local 3D ultrasound echo
signal is phase-matched, cross-spectrum of r.sup.i.sub.1 (l,m,n)
and r.sub.2(l,m,n) is evaluated as
S.sup.i.sub.2,1(l,m,n)=R.sub.2*(l,m,n)R.sup.i.sub.1(l,m,n).
[0138] Cross-spectrum is represented as 5 S 2 , 1 i ( l , m , n ) R
t ' ( l , m , n ) 2 exp { j 2 L u x i ( x , y , z ) x l + j 2 M u y
i ( x , y , z ) y m + j 2 N u z i ( x , y , z ) z n } , ( 4 )
[0139] where 0.ltoreq.l.ltoreq.L-1, 0.ltoreq.m.ltoreq.M-1,
0.ltoreq.n.ltoreq.N-1,
[0140] and then the phase is represented as 6 i ( l , m , n ) = tan
- 1 ( Im [ S 2 , 1 i ( l , m , n ) ] Re [ S 2 , 1 i ( l , m , n ) ]
) , ( 5 )
[0141] where Re[.] and Im[.] respectively represent the real and
imaginary component of ".".
[0142] The least squares method is implemented on the gradient of
the phase eq. (5) weighted with squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m,n).vertline..sup.2(=Re.sup.2[S.sub.2,1.sup.-
i(l,m,n)].sup.2+Im.sup.2[S.sub.2,1.sup.i(l,m,n)]). That is, by
minimizing functional
[0143] error 7 ( u i ( x , y , z ) ) = l , m , n S 2 , 1 i ( l , m
, n ) 2 .times. ( i ( l , m , n ) - u x i ( x , y , z ) ( 2 L x ) l
- u y i ( x , y , z ) ( 2 M y ) m - u z i ( x , y , z ) ( 2 N z ) n
) 2 ( 6 )
[0144] with respect to the 3D residual vector u.sup.i(x,y,z) to be
used to update the i-1 th estimate d.sup.i-1(x,y,z) of the 3D
displacement vector d(x,y,z), the estimate of u.sup.i(x,y,z) is
obtained as
(x,y,z)[=(.sub.z.sup.i(x,y,z), .sub.y.sup.i(x,y,z),
.sub.z.sup.i(x,y,z)).sup.T]. (6-2)
[0145] Concretely, the next simultaneous equations are solved. 8 [
l , m , n S 2 , 1 i ( l , m , n ) 2 ( 2 L x ) l i ( l , m , n ) l ,
m , n S 2 , 1 i ( l , m , n ) 2 ( 2 M y ) m i ( l , m , n ) l , m ,
n S 2 , 1 i ( l , m , n ) 2 ( 2 N z ) n i ( l , m , n ) ] = [ l , m
, n S 2 , 1 i ( l , m , n ) 2 ( 2 L x ) 2 l 2 l , m , n S 2 , 1 i (
l , m , n ) 2 ( 2 L x ) ( 2 M y ) lm l , m , n S 2 , 1 i ( l , m ,
n ) 2 ( 2 L x ) ( 2 N z ) ln l , m , n S 2 , 1 i ( l , m , n ) 2 (
2 L x ) ( 2 M y ) lm l , m , n S 2 , 1 i ( l , m , n ) 2 ( 2 M y )
2 m 2 l , m , n S 2 , 1 i ( l , m , n ) 2 ( 2 M y ) ( 2 N z ) mn l
, m , n S 2 , 1 i ( l , m , n ) 2 ( 2 L x ) ( 2 N z ) ln l , m , n
S 2 , 1 i ( l , m , n ) 2 ( 2 M y ) ( 2 N z ) mn l , m , n S 2 , 1
i ( l , m , n ) 2 ( 2 N z ) 2 n 2 ] .times. [ u x i ( x , y , z ) u
y i ( x , y , z ) u z i ( x , y , z ) ] ( 7 )
[0146] When the 3D displacement vector d (x,y,z) is large, the 3D
residual displacement vector u.sup.i(x,y,z) needs to be estimated
after unwrapping the phase of the cross-spectrum [eq. (3)] in the
frequency domain (l, m, n).
[0147] Alternatively, when the 3D displacement vector d (x,y, z) is
large, by using cross-correlation method (evaluation of the peak
position of the cross-correlation function obtained as 3D inverse
Fourier's transform of the cross-spectrum [eq. (3)]) at the initial
stages during iterative estimation, the 3D residual displacement
vector u.sup.i(x,y,z) can be estimated without unwrapping the phase
of the cross-spectrum [eq. (3)] in the frequency domain.
Specifically, by using the cross-correlation method, x, y, and z
components of the 3D displacement vector are respectively estimated
as integer multiplications of the ultrasound echo sampling
intervals .DELTA.x, .DELTA.y, .DELTA.z. For instance, with respect
to threshold values correTratio or correTdiff, after 9 ; u ^ i ( x
, y , z ) r; ; u ^ i - 1 ( x , y , z ) r; correTratio or ( 8 ) ; u
^ i ( x , y , z ) r; correTdiff ( 8 ' )
[0148] is satisfied with where .parallel.(x,y,z).parallel. and
.parallel..sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimate of the residual
vectors, by using the estimate of the 3D displacement vector
d(x,y,z) as the initial estimate, the 3D residual displacement
vector is estimated from the gradient of the phase of the
cross-spectrum [eq. (3)].
[0149] Empirically it is known that after using cross-correlation
method the conditions
.vertline.u.sup.i.sub.x(x,y,z).vertline..ltoreq..DELTA.x/2- ,
.vertline.u.sup.i.sub.y(x,y,z).vertline..ltoreq..DELTA.y/2,
.vertline.u.sup.i.sub.z(x,y,z).vertline..ltoreq..DELTA.z/2 are
satisfied with. However, for allowing estimation of the 3D residual
displacement vector without unwrapping the phase of the
cross-spectrum,
[0150] the necessary and sufficient condition is 10 u x i ( x , y ,
z ) x + u y i ( x , y , z ) y + u z i ( x , y , z ) z 1 ( 9 )
[0151] or
.vertline.u.sup.i.sub.x(x,y,z).vertline..ltoreq..DELTA.x/3,
.vertline.u.sup.i.sub.y(x,y,z).vertline..ltoreq..DELTA.y/3, and
.vertline.u.sub.z.sup.i(x,y,z).vertline..ltoreq..DELTA.z/3.
(9')
[0152] Therefore, when estimating the gradient of the
cross-spectrum phase after using cross-correlation method, the
acquired ultrasound echo data are thinned out with constant
interval in each direction and the reduced echo data are used such
that the condition (9) or (9') is satisfied with. The iteration
number i increasing, i.e., the magnitude of the 3D residual
displacement vector components u.sup.1.sub.x(x,y,z),
u.sup.i.sub.y(x,y,z), u.sup.i.sub.z(x,y,z) decreasing, the
ultrasound echo data densities are made restored in each direction.
Hence, at initial stages where estimating the gradient of the
cross-spectrum phase, for instance, ultrasound echo signals are
used with one and half times or twice as a long interval as the
original interval in each direction. The densities of the
ultrasound echo signals are made restored in each direction, for
instance, one and half times or twice per iteration.
[0153] Alternatively, when the magnitude of the 3D displacement
vector d(x,y,z) is large, at initial stages, the acquired original
ultrasound echo data can be thinned out with constant interval in
each direction and the reduced echo data can be used such that the
3D residual displacement vector can be estimated without unwrapping
the phase of the cross-spectrum [eq. (3)] in the frequency domain
(l,m,n). Specifically, the acquired original ultrasound echo data
are thinned out with constant interval in each direction and the
reduced ehco data are used such that the condition (9) or (9') is
satisfied with. The iteration number i increasing, i.e., the
magnitude of the 3D residual displacement vector components
u.sup.i.sub.x(x,y,z), u.sup.i.sub.y(x,y,z), u.sup.i.sub.z(x,y,z)
decreasing, the ultrasound echo data densities are made restored in
each direction, for instance, twice per iteration. When the 3D
residual displacement vector components u.sup.i.sub.x(x,y,z),
u.sup.i.sub.y(x,y,z), u.sup.i.sub.z(x,y,z) are estimated, if
neither the condition (9) nor (9') is satisfied with, the values
are truncated such that the conditions are satisfied with.
[0154] The interval of the ultrasound echo signal data are
shortened, for instance, when with respect to threshold values
stepTratio or stepTdiff the condition 11 ; u i ( x , y , z ) r; ; u
i - 1 ( x , y , z ) r; stepTratio ( 10 )
[0155] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.stepTdiff
(10')
[0156] is satisfied with, where .parallel..sup.i(x,y,z).parallel.
and .parallel..sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0157] The condition (10) or (10') can be applied to each direction
component, and in this case the data interval is shorten in each
direction. These are also applied to below-described methods 1-2,
1-3, 1-4, and 1-5.
[0158] (Process 3: Update of the 3D Displacement Vector Estimate of
the Point (x,y,z))
[0159] Thus, the i th estimate of the 3D displacement vector
d(x,y,z) is evaluated as
d.sup.i(x,y,z)=d.sup.i-1(x,y,z)+.sup.i(x,y,z). (11)
[0160] [Process 4: Condition for Heightening the Spatial Resolution
of the 3D Displacement Vector Distribution Measurement (Condition
for Making the Local Space Small)]
[0161] In order to make the spatial resolution high of the 3D
displacement vector distribution measurement, the local space is
made small during iterative estimation. The criteria is
below-described. The processes 1, 2 and 3 are iteratively carried
out till the criteria is satisfied with. When the criteria is
satisfied with, the local space is made small, for instance, the
length of each side is made half. For instance, the criteria is
(12) or (12') with respect to threshold values Tratio or Tdiff. 12
; u i ( x , y , z ) r; ; u i - 1 ( x , y , z ) r; Tratio ( 12 )
[0162] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.Tdiff,
(12')
[0163] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0164] The condition (12) or (12') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0165] (Process 5: Condition for Terminating the Iterative
Estimation of the 3D Displacement Vector of the Point (x y,z))
[0166] Below-described is the criteria for terminating the
iterative estimation of the 3D displacement vector of each point.
The processes 1, 2 and 3 are iteratively carried out till the
criteria is satisfied with. For instance, the criteria is (13) or
(13') with respect to threshold values aboveTratio or aboveTdiff.
13 ; u i ( x , y , z ) r; ; u i - 1 ( x , y , z ) r; aboveTratio (
13 )
[0167] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.aboveTdiff,
(13')
[0168] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0169] (Process 6)
[0170] The 3D displacement vector component distributions are
obtained by carrying out processes 1, 2, 3, 4, and 5 at every point
in the 3D SOI.
[0171] The initial estimate [eq. (1)] of the iterative estimation
of the 3D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0172] [Limitation of Method 1-1]
[0173] The estimate of the 3D displacement vector d(x,y,z) is
iteratively updated at each point (x,y,z) in the 3D SOI. Being
dependent on the SNR of the local 3D echo signals, particularly at
initial stages errors possibly occur when estimating the residual
vector and then phase matching possibly diverges. For instance,
when solving eq. (7) [process 2] or detecting the peak position of
the cross-correlation function [process 2], errors possibly
occur.
[0174] The possibility for divergence of the phase matching is, for
instance, confirmed by the condition (14) or (14') with respect to
the threshold value belowTratio or BelowTdiff. 14 ; u i ( x , y , z
) r; ; u i - 1 ( x , y , z ) r; belowTratio ( 14 )
[0175] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..gtoreq.belowTdiff,
(14')
[0176] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0177] To prevent phase matching (process 1) from diverging, in the
below-described methods 1-2, 1-3, 1-4, and 1-5, by freely using the
condition (14) or (14'), estimation error is reduced of the
residual vector. Thus, even if the SNR of the ultrasound echo
signals are low, accurate 3D displacement vector measurement can be
realized.
[0178] [Method 1-2]
[0179] The flowchart of the method 1-2 is shown in FIG. 11. To
prevent phase matching from diverging at the process 1 of the
method 1-1, estimation error is reduced of the residual vector.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 3D displacement vector measurement can be realized.
[0180] The procedure of iterative estimation is different from that
of the method 1-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0181] (Process 1: Estimation of the 3D Residual Displacement
Vector Distribution)
[0182] Phase matching and estimation of the 3D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the processes 1 and 2 of the method 1-1 are performed
once at every point in the SOI. Thus, the estimate of the 3D
residual vector distribution is obtained [eq. (6-2)].
[0183] (Process 2: Update of the Estimate of the 3D Displacement
Vector Distribution)
[0184] The i-1 th estimate of the 3D displacement vector
distribution is updated using i th estimate of the 3D residual
vector distribution.
d.sup.i(x,y,z)={haeck over (d)}.sup.i-1(x,y,z)+.sup.i(x,y,z)
(15)
[0185] Next, this estimate is 3D low pass filtered or 3D median
filter to yield the estimate of the 3D displacement vector
distribution:
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)]. (16)
[0186] Thus, the estimation error is reduced of the residual vector
compared with process 2 of the method 1-1 [eq. (7)]. Hence, phase
matching of the process 1 of method 1-2 is performed using smoothed
estimate of the 3D displacement vector distribution.
[0187] [Process 3: Condition for Heightening the Spatial Resolution
of the 3D Displacement Vector Distribution Measurement (Condition
for Making the Local Space Small)]
[0188] In order to make the spatial resolution high of the 3D
displacement vector distribution measurement, during iterative
estimation, the local space used for each point is made small, or
the local space used over the SOI is made small.
[0189] The criteria for each point is below-described. The
processes 1 and 2 (method 1-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local space is made small, for instance, the length of each
side is made half. For instance, the criteria is (17) or (17') with
respect to threshold values Tratio or Tdiff. 15 ; u i ( x , y , z )
r; ; u i - 1 ( x , y , z ) r; Tratio ( 17 )
[0190] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.Tdiff,
(17')
[0191] where .sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,z).paralle- l. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0192] The condition (17) or (17') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0193] The criteria over the SOI is below-described. The processes
1 and 2 (method 1-2) are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
space is made small, for instance, the length of each side is made
half. For instance, the criteria is (18) or (18') with respect to
threshold values Tratioroi or Tdiffroi. 16 ( x , y , z ) SOI ; u i
( x , y , z ) r; 2 ( x , y , z ) SOI ; u i - 1 ( x , y , z ) r; 2
Tratioroi ( 18 ) or ( x , y , z ) SOI ; u i ( x , y , z ) - u i - 1
( x , y , z ) r; Tdiffroi , ( 18 ' )
[0194] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0195] The condition (18) or (18') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0196] (Process 4: Condition for Terminating the Iterative
Estimation of the 3D Displacement Vector Distribution)
[0197] Below-described is the criteria for terminating the
iterative estimation of the 3D displacement vector distribution.
The processes 1, 2 and 3 of method 1-2 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(19) or (19') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 17 ( x , y , z ) SOI ; u i ( x , y , z ) r; 2 ( x ,
y , z ) SOI ; u i - 1 ( x , y , z ) r; 2 aboveTratioroi ( 19 ) or (
x , y , z ) SOI ; u i ( x , y , z ) - u i - 1 ( x , y , z ) r;
aboveTdiffroi , ( 19 ' )
[0198] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0199] Final estimate is obtained from eq. (15) or eq. (16).
[0200] The initial estimate [eq. (1)] of the iterative estimation
of the 3D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0201] [Method 1-3]
[0202] The flowchart of the method 1-3 is shown in FIG. 12. To
prevent phase matching from diverging at the process 1 of the
method 1-1, estimation error is reduced of the residual vector.
Possibility of divergence is detected from above-described
condition (14) or (14'), and by effectively utilizing method 1-1
and 1-2, even if the SNR of the ultrasound echo signals are low,
accurate 3D displacement vector measurement can be realized.
[0203] At first, the procedure of iterative estimation is same as
that of the method 1-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0204] Phase matching and estimation of the 3D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the processes 1 and 2 of the method 1-1 are performed
once at every point in the SOI. Thus, the estimate of the 3D
residual vector distribution is obtained [eq. (6-2)].
[0205] During this estimation, if neither condition (14) nor (14')
is satisfied with, the method 1-1 is used. If condition (14) or
(14') is satisfied with at points or spaces, in the process 2 of
the method 1-2, over sufficiently large spaces centered on the
points or spaces, or over the SOI, the estimate d.sup.i(x,y,z) of
the 3D displacement vector d (x,y,z) can be 3D low pass filtered or
3D median filtered as eq. (20).
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (20)
[0206] Thus, the estimation error is reduced of the residual vector
compared with process 2 of the method 1-1 [eq. (7)].
[0207] Thus, iterative estimation is terminated at the process 5 of
the method 1-1 or the process 4 of the method 1-2. Hence, final
estimate is obtained from eq. (11), or eq.(15), or eq. (20).
[0208] The initial estimate [eq. (1)] of the iterative estimation
of the 3D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0209] [Method 1-4]
[0210] The flowchart of the method 1-4 is shown in FIG. 13. To
prevent phase matching from diverging at the process 1 of the
method 1-1, estimation error is reduced of the residual vector.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 3D displacement vector measurement can be realized.
[0211] The procedure of iterative estimation is different from that
of the method 1-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0212] (Process 1: Estimation of the 3D Residual Displacement
Vector Distribution)
[0213] Phase matching and estimation of the 3D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the process 1 of the method 1-1 is performed once at
every point in the SOI.
[0214] To obtain the estimate .sup.i(x,y,z) [=(.sub.x.sup.i(x,y,z),
.sub.y.sup.i(x,y,z), .sub.z.sup.i(x,y,z)).sup.T] of the residual
vector distribution u.sup.i(x,y,z)[=(u.sup.i.sub.x(x,y,z),
u.sup.i.sub.y(x,y,z), u.sub.z(x,y,z)).sup.T],at every point local
3D echo cross-spectrum is evaluated from the 3D Fourier's
transforms of the pre-deformation local 3D ultrasound echo signal
r.sub.1(l,m,n) and phase-matched post-deformation local 3D
ultrasound echo signal r.sup.i.sub.2 (l,m,n). Alternatively, when
pre-deformation local 3D ultrasound echo signal is phase-matched,
at every point cross-spectrum of r.sup.i.sub.1(l,m,n) and
r.sub.2(l,m,n) is evaluated.
[0215] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m,n).vertline..sup.2, where each weight
function is normalized by the power of the cross-spectrum, i.e., 18
l , m , n S 2 , 1 i ( l , m , n ) 2 .
[0216] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 3D residual vector distribution
u.sup.i(x,y,z). 19 error ( u i ) = ; a - F u i r; 2 + 1 i ; u i r;
2 + 2 i ; G u i r; 2 + 3 i ; G T G u i r; 2 + 4 i ; G G T G u i r;
2 ( 21 )
[0217] where
[0218] a: vector comprised of (x,y,z) distribution of the
cross-spectrum phase .THETA..sup.i(l,m,n) weighted with
cross-spectrum .vertline.S.sub.2,1.sup.i(l,m,n).vertline.
normalized by the magnitude of the cross-spectrum 20 l , m , n S 2
, 1 i ( l , m , n ) 2
[0219] evaluated at every point in the 3D SOI.
[0220] F: matrix comprised of (x,y,z) distribution of the Fourier's
coordinate value (l,m,n) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m,n).vertline. normalized by the
magnitude of the cross-spectrum 21 l , m , n S 2 , 1 i ( l , m , n
) 2
[0221] evaluated at every point in the 3D SOI.
[0222] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0223] Gu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the unknown 3D residual vector u.sup.i(x,y,z)
components 22 x u x i ( x , y , z ) , y u x i ( x , y , z ) , z u x
i ( x , y , z ) , x u y i ( x , y , z ) , y u y i ( x , y , z ) , z
u y i ( x , y , z ) , x u z i ( x , y , z ) , y u z i ( x , y , z )
, z u z i ( x , y , z )
[0224] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D Laplacians of the
unknown 3D residual vector u.sup.i(x,y,z) components 23 2 x 2 u x i
( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z
) 2 x 2 u y i ( x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z 2 u y
i ( x , y , z ) 2 x 2 u z i ( x , y , z ) + 2 y 2 u z i ( x , y , z
) + 2 z 2 u z i ( x , y , z )
[0225] GG.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the 3D Laplacians of the unknown 3D residual vector
u.sup.i(x,y,z) components 24 x ( 2 x 2 u x i ( x , y , z ) + 2 y 2
u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z ) ) , y ( 2 x 2 u x i
( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z
) ) , z ( 2 x 2 u x i ( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2
z 2 u x i ( x , y , z ) ) , x ( 2 x 2 u y i ( x , y , z ) + 2 y 2 u
y i ( x , y , z ) + 2 z 2 u y i ( x , y , z ) ) , y ( 2 x 2 u y i (
x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z 2 u y i ( x , y , z )
) , z ( 2 x 2 u y i ( x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z
2 u y i ( x , y , z ) ) , x ( 2 x 2 u z i ( x , y , z ) + 2 y 2 u z
i ( x , y , z ) + 2 z 2 u z i ( x , y , z ) ) , y ( 2 x 2 u z i ( x
, y , z ) + 2 y 2 u z i ( x , y , z ) + 2 z 2 u z i ( x , y , z ) )
, z ( 2 x 2 u z i ( x , y , z ) + 2 y 2 u z i ( x , y , z ) + 2 z 2
u z i ( x , y , z ) ) ,
[0226] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error(u.sup.i) has one minimum value. Thus, by
solving for residual displacement vector distribution
u.sup.i(x,y,z) the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(22)
[0227] estimate .sup.i(x,y,z) [=(.sub.x.sup.i(x,y,z),
.sub.y.sup.i(x,y,z), .sub.z.sup.i(x,y,z)).sup.T] of the residual
vector distribution u.sup.i(x,y,z) [=(u.sup.i.sub.x(x,y,z),
u.sup.i.sub.y (x,y,z), u.sup.i.sub.z(x,y,z)).sup.T] is stably
obtained. Thus, estimation error is reduced of the residual
vector.
[0228] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement vector, direction of the unknown displacement
component, direction of the partial derivative, etc.
[0229] (Process 2: Update of the Estimate of the 3D Displacement
Vector Distribution)
[0230] The i-1 th estimate of the 3D displacement vector
distribution is updated using i th estimate of the 3D residual
vector distribution.
d.sup.i(x,y,z)={haeck over (d)}.sup.i-1(x,y,z)+.sup.i(x,y,z)
(23)
[0231] Freely, this estimate can be 3D low pass filtered or 3D
median filter to yield the estimate of the 3D displacement vector
distribution.
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (24)
[0232] Hence, phase matching of the process 1 of method 1-4 is
performed using the 3D residual vector data u.sup.i(x,y,z) obtained
from eq. (22), or the 3D vector data d.sup.i (x,y,z) obtained from
eq. (23), or smoothed estimate obtained from eq. (24).
[0233] [Process 3: Condition for Heightening the Spatial Resolution
of the 3D Displacement Vector Distribution Measurement (Condition
for Making the Local Space Small)]
[0234] In order to make the spatial resolution high of the 3D
displacement vector distribution measurement, during iterative
estimation, the local space used for each point is made small, or
the local space used over the SOI is made small.
[0235] The criteria for each point is below-described. The
processes 1 and 2 of method 1-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local space is made small, for instance, the length of
each side is made half. For instance, the criteria is (25) or (25')
with respect to threshold values Tratio or Tdiff. 25 ; u ^ i ( x ,
y , z ) r; ; u ^ i - 1 ( x , y , z ) r; Tratio ( 25 )
[0236] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).ltoreq.Tdiff, (25')
[0237] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0238] The condition (25) or (25') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0239] The criteria over the SOI is below-described. The processes
1 and 2 of method 1-4 are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
space is made small, for instance, the length of each side is made
half. For instance, the criteria is (26) or (26') with respect to
threshold values Tratioroi or Tdiffroi. 26 ( x , y , z ) SOI ; u ^
i ( x , y , z ) r; 2 ( x , y , z ) SOI ; u ^ i - 1 ( x , y , z ) r;
2 Tratioroi or ( 26 ) ( x , y , z ) SOI ; u ^ i ( x , y , z ) - u ^
i - 1 ( x , y , z ) r; Tdiffroi , ( 26 ' )
[0240] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0241] The condition (26) or (26') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0242] (Process 4: Condition for Terminating the Iterative
Estimation of the 3D Displacement Vector Distribution)
[0243] Below-described is the criteria for terminating the
iterative estimation of the 3D displacement vector distribution.
The processes 1, 2 and 3 of method 1-4 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(27) or (27') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 27 ( x , y , z ) SOI ; u ^ i ( x , y , z ) r; 2 ( x
, y , z ) SOI ; u ^ i - 1 ( x , y , z ) r; 2 above Tratioroi or (
27 ) ( x , y , z ) SOI ; u ^ i ( x , y , z ) - u ^ i - 1 ( x , y ,
z ) r; aboveTdiffroi , ( 27 ' )
[0244] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0245] Final estimate is obtained from eq. (23) or eq. (24).
[0246] The initial estimate [eq. (1)] of the iterative estimation
of the 3D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0247] [Method 1-5]
[0248] The flowchart of the method 1-5 is shown in FIG. 14. To
prevent phase matching from diverging at the process 1 of the
method 1-1, estimation error is reduced of the residual vector.
Possibility of divergence is detected from above-described
condition (14) or (14'), and by effectively utilizing method 1-1
and 1-4, even if the SNR of the ultrasound echo signals are low,
accurate 3D displacement vector measurement can be realized.
[0249] At first, the procedure of iterative estimation is same as
that of the method 1-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0250] Phase matching and estimation of the 3D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the process 1 of the method 1-1 is performed once at
every point in the SOI. Moreover, using the regularization method,
stably the estimate of the 3D residual vector distribution is
obtained.
[0251] During this estimation, if neither condition (14) nor (14')
is satisfied with, the method 1-1 is used. If condition (14) or
(14') is satisfied with at points or spaces, in the process 2 of
the method 1-4, over sufficiently large spaces centered on the
points or spaces, or over the SOI, the estimate d.sup.i(x,y,z) of
the 3D displacement vector d (x,y,z) can be 3D low pass filtered or
3D median filtered as eq. (28).
{haeck over (d)}.sup.i(x,y,z)=LPF [d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (28)
[0252] Thus, the estimation error is reduced of the residual
vector.
[0253] Iterative estimation is terminated at the process 5 of the
method 1-1 or the process 4 of the method 1-4. Hence, final
estimate is obtained from eq. (11), or eq.(23), or eq.(28).
[0254] The initial estimate [eq. (1)] of the iterative estimation
of the 3D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0255] (II) Method 2: Measurement of 2D Displacement Vector
Component Distribution in 2D ROI
[0256] 2D displacement vector distribution can be measured in 2D
ROI 7 in the Cartesian coordinate system. 2D ultrasound echo
signals r.sub.1 (x,y) and r.sub.2 (x,y) are respectively acquired
under pre-deformation and post-deformation. These echo signals are
processed by the below-described methods 2-1, 2-2, 2-3, 2-4, and
2-5. That is, as shown in FIG. 15, local region is set at each
point in the pre- and post-deformation 2D echo signal, and as shown
in FIG. 16, the corresponding local region is iteratively searched
for in the ROI 7 using the local phase characteristics as the
index. In this searching scheme, the estimated residual
displacement vector is used to update the previously estimated
displacement vector. When the estimated residual displacement
vector satisfies with prescribed condition, the local region size
is made small (FIG. 17). Thus, accurate 2D displacement vector
measurement is realized. Here, sampling intervals are .DELTA.x and
.DELTA.y respectively in the x and y-axes.
[0257] [Method 2-1]
[0258] The procedure of the method 2-1 is shown in FIG. 10. The
processes from 1 to 5 yields 2D displacement vector
d(x,y)[=(dx(x,y), dy(x,y)).sup.T] of arbitrary point (x,y) in 2D
ROI from pre- and post-deformation local 2D echo signals r.sub.1
(l,m) and r.sub.2(l,m) [0.ltoreq.l.ltoreq.L-1,
0.ltoreq.m.ltoreq.M-1] centered on (x,y) of pre- and
post-deformation 2D echo signals r.sub.1(x,y) and r.sub.2(x,y). L
and M must be determined such that .DELTA.xL and .DELTA.yM are
respectively at least 4 times longer than corresponding
displacement components .vertline.dx(x,y).vertline. and
.vertline.dy(x,y).vertline..
[0259] (Process 1: Phase Matching at the Point (x,y))
[0260] Phase matching is performed to obtain i-th estimate
d.sup.i(x,y) [=(d.sup.ix(x,y), d.sup.iy(x,y)).sup.T] of the 2D
displacement vector d(x,y)[=(dx(x,y), dy(x,y)).sup.T].
[0261] Searching region is set in the post-deformation echo signal
space r.sub.2 (x,y), being centered on the local region
[0.ltoreq.1.ltoreq.L-1, 0.ltoreq.m.ltoreq.M-1] centered on (x,y)
and being twice longer than the corresponding length, in order to
update the i-1 th estimate d.sup.i-1(x,y) [=(d.sup.i-1 x (x,y),
d.sup.i-1y (x, y)).sup.T] of the 2D displacement vector
d(x,y)[=(dx(x,y), dy(x,y).sup.T), where
d.sup.0(x,y)={haeck over (d)}(x,y). (29)
[0262] The phase of the post-deformation local echo signal is
matched to pre-deformation local echo signal by multiplying 28 exp
{ j 2 L d x i - 1 ( x , y ) x l + j 2 M d y i - 1 ( x , y ) y m } (
30 )
[0263] to 2D Fourier's transform of this searching region echo
signal r'.sub.2(l,m) [0.ltoreq.l.ltoreq.2L-1,
0.ltoreq.m.ltoreq.2M-1]using i-th estimate d.sup.i-1(x,y), or by
multiplying 29 exp { j 2 L u ^ x i - 1 ( x , y ) x l + j 2 M u ^ y
i - 1 ( x , y ) y m } , ( 30 ' )
[0264] to 2D Fourier's transform of the i-1 th phase-matched
searching region echo signal r'.sup.i-1.sub.2 (l,m) using the
estimate .sub.i-1(x,y) [=(.sub.x.sup.i-1(x,y),
.sub.y.sup.i-1(x,y)).sup.T] [.sup.0(x,y)=0 (zero vector)] of the
vector u.sup.i-1 (x,y) [=(u.sup.i-1x(X,Y),
u.sup.i-1.sub.y(x,y)).sup.T].
[0265] By carrying out inverse Fourier's transform of this product,
post-deformation echo signal r.sup.i.sub.2 (l,m) is obtained at the
center of the searching region echo signal r'.sup.i.sub.2 (l,m)
which is used at i-th stage to estimate 2D displacement vector
d(x,y)[=(dx(x,y), dy(x,y)).sup.T].
[0266] Alternatively, the phase of the pre-deformation local echo
signal can be matched to post-deformation local echo signal in a
similar way. That is, 2D Fourier's transform of the searching
region echo signal r'.sub.1(l,m) [0.ltoreq.l.ltoreq.2L-1,
0.ltoreq.m.ltoreq.2M-1] centered on the point (x,y) in the
pre-deformation echo signal region is multiplied with 30 exp { j 2
L d x i - 1 ( x , y ) x l + j 2 M d y i - 1 ( x , y ) y m } , ( 30
"' )
[0267] or 2D Fourier's transform of the i-1 th phase-matched
searching region echo signal r'.sup.i-1.sub.1 (l,m) is multiplied
with 31 exp { - j 2 L u x i - 1 ( x , y ) x l - j 2 M u y i - 1 ( x
, y ) y m } . ( 30 ''' )
[0268] (Process 2: Estimation of 2D Residual Displacement Vector at
the Point (x,y))
[0269] Local 2D echo cross-spectrum is evaluated from the 2D
Fourier's transforms of the pre-deformation local 2D ultrasound
echo signal r l(l,m) and phase-matched post-deformation local 2D
ultrasound echo signal r.sup.i.sub.2 (l,m)
S.sup.i.sub.2,1(l,m)=R.sub.2.sup.i*(l,m)R.sub.1(l,m), (31)
[0270] where * denotes conjugate.
[0271] Alternatively, when pre-deformation local 2D ultrasound echo
signal is phase-matched, cross-spectrum of r.sup.i.sub.1(l,m) and
r.sub.2(l,m) is evaluated as
S.sup.i.sub.2,1(l,m)=R.sub.2*(l,m)R.sup.i.sub.1(l,m)
[0272] Cross-spectrum is represented as 32 S 2 , 1 i ( l , m ) R 1
i ( l , m ) 2 exp { j 2 L u x i ( x , y ) x l + j 2 M u y i ( x , y
) y } , ( 32 )
[0273] where 0.ltoreq.l.ltoreq.L-1, 0.ltoreq.m.ltoreq.M-1,
[0274] and then the phase is represented as 33 i ( l , m ) = tan -
1 ( Im [ S 2 , 1 i ( l , m ) ] Re [ S 2 , 1 i ( l , m ) ] ) , ( 33
)
[0275] where Re[.] and Im[.] respectively represent the real and
imaginary component of ".".
[0276] The least squares method is implemented on the gradient of
the phase eq. (33) weighted with squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m).vertline..sup.2(=Re.sup.2[S.sub.2,1.sup.i(l-
,m)].sup.2+Im.sup.2[S.sub.2,1.sup.i(l,mr)]) .
[0277] That is, by minimizing functional: 34 error ( u i ( x , y )
) = l , m S 2 , 1 i ( l , m ) 2 ( i ( l , m ) - u x i ( x , y ) ( 2
L x ) l - u y i ( x , y ) ( 2 M y ) m ) 2 ( 34 )
[0278] with respect to the 2D residual vector u.sup.i(x,y) to be
used to update the i-1 th estimate d.sup.i-1(x,y) of the 2D
displacement vector d(x,y), the estimate of u.sup.i(x,y) is
obtained as
.sup.i(x,y)[=(.sub.x.sup.i(x,y), .sub.y.sup.i(x,y)).sup.T].
[0279] Concretely, the next simultaneous equations are solved. 35 [
l , m S 2 , 1 i ( l , m ) 2 ( 2 L x ) l i ( l , m ) l , m S 2 , 1 i
( l , m ) 2 ( 2 M y ) m i ( l , m ) ] = [ l , m S 2 , 1 i ( l , m )
2 ( 2 L x ) 2 l 2 l , m S 2 , 1 i ( l , m ) 2 ( 2 L x ) ( 2 M y ) l
m l , m S 2 , 1 i ( l , m ) 2 ( 2 L x ) ( 2 M y ) l m l , m S 2 , 1
i ( l , m ) 2 ( 2 M y ) 2 m 2 ] .times. [ u x i ( x , y ) u y i ( x
, y ) ] ( 35 )
[0280] When the 2D displacement vector d (x,y) is large, the 2D
residual displacement vector u.sup.i (x,y) needs to be estimated
after unwrapping the phase of the cross-spectrum [eq. (31)] in the
frequency domain (l,m).
[0281] Alternatively, when the 2D displacement vector d (x,y) is
large, by using cross-correlation method (evaluation of the peak
position of the cross-correlation function obtained as 2D inverse
Fourier's transform of the cross-spectrum [eq. (31)]) at the
initial stages during iterative estimation, the 2D residual
displacement vector u.sup.i (x,y) can be estimated without
unwrapping the phase of the cross-spectrum [eq. (31)] in the
frequency domain. Specifically, by using the cross-correlation
method, x and y components of the 2D displacement vector are
respectively estimated as integer multiplications of the ultrasound
echo sampling intervals .DELTA.x, .DELTA.y. For instance, with
respect to threshold values correTratio or correTdiff, after 36 ; u
i ( x , y ) r; ; u i - 1 ( x , y ) r; correTratio ( 36 )
[0282] or
.parallel..sup.i(x,y).parallel..ltoreq.correTdiff (36')
[0283] is satisfied with where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors, by using the estimate of the 2D displacement vector d(x,y)
as the initial estimate, the 2D residual displacement vector is
estimated from the gradient of the phase of the cross-spectrum [eq.
(31)].
[0284] Empirically it is known that after using cross-correlation
method the conditions
.vertline.u.sup.i.sub.x(x,y).vertline..ltoreq..DELTA.x/2,
.vertline.u.sup.i.sub.y(x,y).vertline..ltoreq..DELTA.y/2 are
satisfied with. Then, the necessary and sufficient condition for
allowing estimation of the 2D residual displacement vector without
unwrapping the phase of the cross-spectrum 37 u x i ( x , y ) x + u
y i ( x , y ) y 1 ( 37 )
[0285] is satisfied with.
[0286] Alternatively, when the magnitude of the 2D displacement
vector d (x,y) is large, at initial stages, the acquired original
ultrasound echo data can be thinned out with constant interval in
each direction and the reduced echo data can be used such that the
2D residual displacement vector can be estimated without unwrapping
the phase of the cross-spectrum [eq. (31)] in the frequency domain
(l,m). Specifically, the acquired original ultrasound echo data are
thinned out with constant interval in each direction and the
reduced ehco data are used such that the condition (37) or (37') is
satisfied with.
.vertline..sub.x.sup.i(x,y).vertline..ltoreq..DELTA.x/2 and
.vertline.u.sub.y.sup.i(x,y).vertline..ltoreq..DELTA.y/2. (37')
[0287] The iteration number i increasing, i.e., the magnitude of
the 2D residual displacement vector components u.sup.i.sub.x(x,y),
u.sup.i.sub.y(x,y) decreasing, the ultrasound echo data densities
are made restored in each direction, for instance, twice per
iteration.
[0288] The interval of the ultrasound echo signal data are
shortened, for instance, when with respect to threshold values
stepTratio or stepTdiff the condition 38 ; u i ( x , y ) r; ; u i -
1 ( x , y ) r; stepTratio ( 38 )
[0289] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..ltoreq.stepTdiff
(38')
[0290] is satisfied with, where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0291] The condition (38) or (38') can be applied to each direction
component, and in this case the data interval is shorten in each
direction. These are also applied to below-described methods 2-2,
2-3, 2-4, and 2-5.
[0292] (Process 3: Update of the 2D Displacement Vector Estimate of
the Point (x,y))
[0293] Thus, the i th estimate of the 2D displacement vector d(x,y)
is evaluated as
d.sup.i(x,y)=d.sup.i-1(x,y)+.sup.i(x,y). (39)
[0294] [Process 4: Condition for Heightening the Spatial Resolution
of the 2D Displacement Vector Distribution Measurement (Condition
for Making the Local Region Small)]
[0295] In order to make the spatial resolution high of the 2D
displacement vector distribution measurement, the local region is
made small during iterative estimation. The criteria is
below-described. The processes 1, 2 and 3 are iteratively carried
out till the criteria is satisfied with. When the criteria is
satisfied with, the local region is made small, for instance, the
length of each side is made half. For instance, the criteria is
(40) or (40') with respect to threshold values Tratio or Tdiff. 39
; u i ( x , y ) r; ; u i - 1 ( x , y ) r; Tratio ( 40 )
[0296] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..ltoreq.Tdiff,
(40')
[0297] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0298] The condition (40) or (40') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0299] (Process 5: Condition for Terminating the Iterative
Estimation of the 2D Displacement Vector of the Point (x,y))
[0300] Below-described is the criteria for terminating the
iterative estimation of the 2D displacement vector of each point.
The processes 1, 2 and 3 are iteratively carried out till the
criteria is satisfied with. For instance, the criteria is (41) or
(41') with respect to threshold values aboveTratio or aboveTdiff.
40 ; u i ( x , y ) r; ; u i - 1 ( x , y ) r; aboveTratio ( 41 )
[0301] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..ltoreq.aboveTdiff,
(41')
[0302] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0303] (Process 6)
[0304] The 2D displacement vector component distributions are
obtained by carrying out processes 1, 2, 3, 4, and 5 at every point
in the 2D ROI.
[0305] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0306] [Limitation of Method 2-1]
[0307] The estimate of the 2D displacement vector d(x,y) is
iteratively updated at each point (x,y) in the 2D ROI. Being
dependent on the SNR of the local 2D echo signals, particularly at
initial stages errors possibly occur when estimating the residual
vector and then phase matching possibly diverges. For instance,
when solving eq. (35) [process 2] or detecting the peak position of
the cross-correlation function [process 2], errors possibly
occur.
[0308] The possibility for divergence of the phase matching is, for
instance, confirmed by the condition (42) or (42') with respect to
the threshold value belowTratio or BelowTdiff. 41 ; u ^ i ( x , y )
r; ; u ^ i - 1 ( x , y ) r; belowTratio ( 42 )
[0309] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..gtoreq.belowTdiff,
(42')
[0310] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0311] To prevent phase matching (process 1) from diverging, in the
below-described methods 2-2, 2-3, 2-4, and 2-5, by freely using the
condition (42) or (42'), estimation error is reduced of the
residual vector. Thus, even if the SNR of the ultrasound echo
signals are low, accurate 2D displacement vector measurement can be
realized.
[0312] [Method 2-2]
[0313] The flowchart of the method 2-2 is shown in FIG. 11. To
prevent phase matching from diverging at the process 1 of the
method 2-1, estimation error is reduced of the residual vector.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0314] The procedure of iterative estimation is different from that
of the method 2-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0315] (Process 1: Estimation of the 2D Residual Displacement
Vector Distribution)
[0316] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y) in the 2D
ROI. That is, the processes 1 and 2 of the method 2-1 are performed
once at every point in the ROI. Thus, the estimate of the 2D
residual vector distribution is obtained.
[0317] (Process 2: Update of the Estimate of the 2D Displacement
Vector Distribution)
[0318] The i-1 th estimate of the 2D displacement vector
distribution is updated using i th estimate of the 2D residual
vector distribution.
d.sup.i(x,y)={haeck over (d)}.sup.i-1(x,y)+.sup.i(x,y) (43)
[0319] Next, this estimate is 2D low pass filtered or 2D median
filter to yield the estimate of the 2D displacement vector
distribution:
{haeck over (d)}.sup.i(x,y)=LPF[d.sup.i(x,y)], or {haeck over
(d)}.sup.i(x,y)=MED[d.sup.i(x,y)]. (44)
[0320] Thus, the estimation error is reduced of the residual vector
compared with process 2 of the method 2-1 [eq. (35)]. Hence, phase
matching of the process 1 of method 2-2 is performed using smoothed
estimate of the 2D displacement vector distribution.
[0321] [Process 3: Condition for Heightening the Spatial Resolution
of the 2D Displacement Vector Distribution Measurement (Condition
for Making the Local Region Small)]
[0322] In order to make the spatial resolution high of the 2D
displacement vector distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the ROI is made small.
[0323] The criteria for each point is below-described. The
processes 1 and 2 (method 2-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of each
side is made half. For instance, the criteria is (45) or (45') with
respect to threshold values Tratio or Tdiff. 42 ; u ^ i ( x , y )
r; ; u ^ i - 1 ( x , y ) r; Tratio ( 45 )
[0324] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..ltoreq.Tdiff,
(45')
[0325] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0326] The condition (45) or (45') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0327] The criteria over the ROI is below-described. The processes
1 and 2 (method 2-2) are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
region is made small, for instance, the length of each side is made
half. For instance, the criteria is (46) or (46') with respect to
threshold values Tratioroi or Tdiffroi. 43 ( x , y ) ROI ; u ^ i (
x , y ) r; 2 ( x , y ) ROI ; u ^ i - 1 ( x , y ) r; 2 Tratioroi or
( 46 ) ( x , y ) ROI ; u ^ i ( x , y ) - u ^ i - 1 ( x , y ) r;
Tdiffroi ( 46 ' )
[0328] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0329] The condition (46) or (46') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0330] (Process 4: Condition for Terminating the Iterative
Estimation of the 2D Displacement Vector Distribution)
[0331] Below-described is the criteria for terminating the
iterative estimation of the 2D displacement vector distribution.
The processes 1, 2 and 3 of method 2-2 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(47) or (47') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 44 ( x , y ) ROI ; u ^ i ( x , y ) r; 2 ( x , y )
ROI ; u ^ i - 1 ( x , y ) r; 2 aboveTratioroi or ( 47 ) ( x , y )
ROI ; u ^ i ( x , y ) - u ^ i - 1 ( x , y ) r; aboveTdiffroi , ( 47
' )
[0332] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0333] Final estimate is obtained from eq. (43) or eq. (44).
[0334] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0335] [Method 2-3]
[0336] The flowchart of the method 2-3 is shown in FIG. 12. To
prevent phase matching from diverging at the process 1 of the
method 2-1, estimation error is reduced of the residual vector.
Possibility of divergence is detected from above-described
condition (42) or (42'), and by effectively utilizing method 2-1
and 2-2, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0337] At first, the procedure of iterative estimation is same as
that of the method 2-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0338] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y) in the 2D
ROI. That is, the processes 1 and 2 of the method 2-1 are performed
once at every point in the ROI. Thus, the estimate of the 2D
residual vector distribution is obtained.
[0339] During this estimation, if neither condition (42) nor (42')
is satisfied with, the method 2-1 is used. If condition (42) or
(42') is satisfied with at points or regions, in the process 2 of
the method 2-2, over sufficiently large regions centered on the
points or regions, or over the ROI, the estimate d.sup.i(x,y) of
the 2D displacement vector d (x,y) can be 2D low pass filtered or
2D median filtered as eq. (48).
{haeck over (d)}.sup.i(x,y)=LPF[d.sup.i(x,y)], or {haeck over
(d)}.sup.i(x,y)=MED[d.sup.i(x,y)] (48)
[0340] Thus, the estimation error is reduced of the residual vector
compared with process 2 of the method 2-1 [eq. (35)].
[0341] Thus, iterative estimation is terminated at the process 5 of
the method 2-1 or the process 4 of the method 2-2. Hence, final
estimate is obtained from eq. (39), or eq.(43), or eq.(48).
[0342] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0343] [Method 2-4]
[0344] The flowchart of the method 2-4 is shown in FIG. 13. To
prevent phase matching from diverging at the process 1 of the
method 2-1, estimation error is reduced of the residual vector.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0345] The procedure of iterative estimation is different from that
of the method 2-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0346] (Process 1: Estimation of the 2D Residual Displacement
Vector Distribution)
[0347] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y) in the 2D
ROI. That is, the process 1 of the method 2-1 is performed once at
every point in the ROI.
[0348] To obtain the estimate .sup.i(x,y) [=(.sub.x.sup.i(x,y),
(x,y)).sup.T] of the residual vector distribution u.sup.i
(x,y)[=(u.sup.i.sub.x(x,y), u.sup.i.sub.y(x,y)).sup.T], at every
point local 2D echo cross-spectrum is evaluated from the 2D
Fourier's transforms of the pre-deformation local 2D ultrasound
echo signal r.sub.1(l,m) and phase-matched post-deformation local
2D ultrasound echo signal r.sup.i.sub.2 (l,m). Alternatively, when
pre-deformation local 2D ultrasound echo signal is phase-matched,
at every point cross-spectrum of r.sup.i.sub.1(l,m) and
r.sub.2(l,m) is evaluated.
[0349] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m).vertline..sup.2 where each weight
function is normalized by the power of the cross-spectrum, 45 i . e
. , l , m S 2 , 1 i ( l , m ) 2 .
[0350] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 2D residual vector distribution
u.sup.i(x,y).
error (u.sup.i)=.vertline..vertline.a-F
u.sup.i.vertline..vertline..sup.2+-
.alpha..sub.1i.vertline..vertline.u.sup.i.vertline..vertline..sup.2+.alpha-
..sub.2i.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2+.alpha..sub-
.3i.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2+.alpha..s-
ub.4i.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2
(49)
[0351] where
[0352] a: vector comprised of (x,y) distribution of the
cross-spectrum phase .THETA..sup.i(l,m) weighted with
cross-spectrum .vertline.S.sub.2,1.sup.i(l,m).vertline. normalized
by the magnitude of the cross-spectrum 46 l , m S 2 , 1 i ( l , m )
2
[0353] evaluated at every point in the 2D ROI.
[0354] F: matrix comprised of (x,y) distribution of the Fourier's
coordinate value (l,m) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m).vertline. normalized by the
magnitude of the cross-spectrum 47 l , m S 2 , 1 i ( l , m ) 2
[0355] evaluated at every point in the 2D ROI.
[0356] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0357] Gu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D gradient
components of the unknown 2D residual vector u.sup.i(x,y)
components 48 x u x i ( x , y ) , y u x i ( x , y ) , x u y i ( x ,
y ) , y u y i ( x , y )
[0358] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D Laplacians of the
unknown 2D residual vector u.sup.i(x,y) components 49 2 x 2 u x i (
x , y ) , 2 y 2 u x i ( x , y ) 2 x 2 u y i ( x , y ) , 2 y 2 u y i
( x , y )
[0359] GG.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D gradient
components of the 2D Laplacians of the unknown 2D residual vector
u.sup.i(x,y) components 50 x ( 2 x 2 u x i ( x , y ) + 2 y 2 u x i
( x , y ) ) , y ( 2 x 2 u x i ( x , y ) + 2 y 2 u x i ( x , y ) ) ,
x ( 2 x 2 u y i ( x , y ) + 2 y 2 u y i ( x , y ) ) , y ( 2 x 2 u y
i ( x , y ) + 2 y 2 u y i ( x , y ) )
[0360] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error(u.sup.i) has one minimum value. Thus, by
solving for residual displacement vector distribution u.sup.i(x,y)
the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(50)
[0361] estimate .sup.i(x,y)[=(.sub.x.sup.i(x,y),
.sub.y.sup.i(x,y)).sup.T] of the residual vector distribution
u.sup.i(x,y) [=(u.sup.i.sub.x(x,y), u.sup.i.sub.y(x,y)).sup.T] is
stably obtained. Thus, estimation error is reduced of the residual
vector.
[0362] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement vector, direction of the unknown displacement
component, direction of the partial derivative, etc.
[0363] (Process 2: Update of the Estimate of the 2D Displacement
Vector Distribution)
[0364] The i-1 th estimate of the 2D displacement vector
distribution is updated using i th estimate of the 2D residual
vector distribution.
d.sup.i(x,y)={haeck over (d)}.sup.i-1)+.sup.i(x,y) (51)
[0365] Freely, this estimate can be 2D low pass filtered or 2D
median filter to yield the estimate of the 2D displacement vector
distribution.
{haeck over (d)}.sup.i(x,y)=LPF[d.sup.i(x,y)], or {haeck over
(d)}.sup.i(x,y)=MED[d.sup.i(x,y)] (52)
[0366] Hence, phase matching of the process 1 of method 2-4 is
performed using the 2D residual vector data u.sup.i(x,y) obtained
from eq. (50), or the 2D vector data d.sup.i(x,y) obtained from eq.
(51), or smoothed estimate obtained from eq. (52).
[0367] [Process 3: Condition for Heightening the Spatial Resolution
of the 2D Displacement Vector Distribution Measurement (Condition
for Making the Local Region Small)]
[0368] In order to make the spatial resolution high of the 2D
displacement vector distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the ROI is made small.
[0369] The criteria for each point is below-described. The
processes 1 and 2 of method 2-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
each side is made half. For instance, the criteria is (25) or (25')
with respect to threshold values Tratio or Tdiff. 51 ; u ^ i ( x ,
y ) r; ; u ^ i - 1 ( x , y ) r; Tratio ( 53 )
[0370] or
.parallel..sup.i(x,y)-.sup.i-1(x,y).parallel..ltoreq.Tdiff,
(53')
[0371] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0372] The condition (53) or (53') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0373] The criteria over the ROI is below-described. The processes
1 and 2 of method 2-4 are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
region is made small, for instance, the length of each side is made
half. For instance, the criteria is (54) or (54') with respect to
threshold values Tratioroi or Tdiffroi. 52 ( x , y ) ROI ; u ^ i (
x , y ) r; ( x , y ) ROI ; u ^ i - 1 ( x , y ) r; Tratioroi or ( 54
) ( x , y ) ROI ; u ^ i ( x , y ) - u ^ i - 1 ( x , y ) r; Tdiffroi
, ( 54 ' )
[0374] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0375] The condition (54) or (54') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0376] (Process 4: Condition for Terminating the Iterative
Estimation of the 2D Displacement Vector Distribution)
[0377] Below-described is the criteria for terminating the
iterative estimation of the 2D displacement vector distribution.
The processes 1, 2 and 3 of method 2-4 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(55) or (55') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 53 ( x , y ) ROI ; u ^ i ( x , y ) r; ( x , y ) ROI
; u ^ i - 1 ( x , y ) r; aboveTratioroi or ( 55 ) ( x , y ) ROI ; u
^ i ( x , y ) - u ^ i - 1 ( x , y ) r; aboveTdiffroi , ( 55 ' )
[0378] where .parallel..sup.i(x,y).parallel. and
.parallel..sup.i-1(x,y).p- arallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0379] Final estimate is obtained from eq. (51) or eq. (52).
[0380] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0381] [Method 2-5]
[0382] The flowchart of the method 2-5 is shown in FIG. 14. To
prevent phase matching from diverging at the process 1 of the
method 2-1, estimation error is reduced of the residual vector.
Possibility of divergence is detected from above-described
condition (42) or (42'), and by effectively utilizing method 2-1
and 2-4, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0383] At first, the procedure of iterative estimation is same as
that of the method 2-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0384] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y) in the 2D
ROI. That is, the process 1 of the method 2-1 is performed once at
every point in the ROI. Moreover, using the regularization method,
stably the estimate of the 2D residual vector distribution is
obtained.
[0385] i-1 th estimate {haeck over (d)}.sup.i-1(x,y) of 2D
displacement vector distribution d(x,y).
[0386] i th estimate .sup.i(x,y) of 2D residual vector distribution
u.sup.i(x, y).
[0387] During this estimation, if neither condition (42) nor (42')
is satisfied with, the method 2-1 is used. If condition (42) or
(42') is satisfied with at points or regions, in the process 2 of
the method 2-4, over sufficiently large regions centered on the
points or regions, or over the ROI, the estimate d.sup.i(x,y) of
the 2D displacement vector d (x,y) can be 2D low pass filtered or
2D median filtered as eq. (56).
{haeck over (d)}.sup.i(x,y)=LPF[d.sup.i(x,y)], or {haeck over
(d)}.sup.i(x,y)=MED[d.sup.i(x,y)] (56)
[0388] Thus, the estimation error is reduced of the residual
vector.
[0389] Iterative estimation is terminated at the process 5 of the
method 2-1 or the process 4.of the method 2-4. Hence, final
estimate is obtained from eq. (39), or eq.(51), or eq.(56).
[0390] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0391] (III) Method 3: Measurement of 1D (One Direction)
Displacement Component Distribution in 1D ROI
[0392] 1D displacement component distribution can be measured in 1D
ROI 7 in the Cartesian coordinate system. 1D ultrasound echo
signals r.sub.1 (x) and r.sub.2 (x) are respectively acquired under
pre-deformation and post-deformation. These echo signals are
processed by the below-described methods 3-1, 3-2, 3-3, 3-4, and
3-5. That is, as shown in FIG. 18, local region is set at each
point in the pre- and post-deformation 1D echo signal, and as shown
in FIG. 19, the corresponding local region is iteratively searched
for in the ROI 7 using the local phase characteristics as the
index. In this searching scheme, the estimated residual
displacement component is used to update the previously estimated
displacement component. When the estimated residual displacement
component satisfies with prescribed condition, the local region
size is made small (FIG. 20). Thus, accurate 1D displacement
component measurement is realized. Here, sampling interval is
.DELTA.x in the x-axis.
[0393] [Method 3-1]
[0394] The procedure of the method 3-1 is shown in FIG. 10. The
processes from 1 to 5 yields 1D displacement component dx(x) of
arbitrary point x in 1D ROI from pre- and post-deformation local 1D
echo signals r.sub.1(l) and r.sub.2(l) [0.ltoreq.l.ltoreq.L-1]
centered on x of pre- and post-deformation 1D echo signals
r.sub.1(x) and r.sub.2(x). L must be determined such that .DELTA.xL
is at least 4 times longer than the displacement component
.vertline.dx(x).vertline..
[0395] (Process 1: Phase Matching at the Point x)
[0396] Phase matching is performed to obtain i-th estimate
d.sup.ix(x) of the 1D displacement component dx(x).
[0397] Searching region is set in the post-deformation echo signal
space r,(x), being centered on the local region
[0.ltoreq.l.ltoreq.L-1] centered on x and being twice longer than
the local region length, in order to update the i-1 th estimate
d.sup.i-1 x(x) of the 1D displacement component dx(x), where
dx.sup.0(x)={haeck over (d)}x(x). (57)
[0398] The phase of the post-deformation local echo signal is
matched to pre-deformation local echo signal by multiplying 54 exp
{ j 2 L d x i - l ( x ) x l } ( 58 )
[0399] to 1D Fourier's transform of this searching region echo
signal r'.sub.2(l) [0.ltoreq.l.ltoreq.2L-1] using i-th estimate
dx.sup.i-1 (x), or by multiplying 55 exp { j 2 L u ^ x i - l ( x )
x l } ( 58 ' )
[0400] to 1D Fourier's transform of the i-1 th phase-matched
searching region echo signal r'.sup.i-1.sub.2 (l) using the
estimate .sub.x.sup.i-1(x) [.sub.x.sup.0(x)=0 (zero)]of the
component u.sup.i-1.sub.x(x) By carrying out inverse Fourier's
transform of this product, post-deformation echo signal
r.sup.i.sub.2 (l) is obtained at the center of the searching region
echo signal r'.sup.i.sub.2 (l), which is used at i-th stage to
estimate 1D displacement component dx(x).
[0401] Alternatively, the phase of the pre-deformation local echo
signal can be matched to post-deformation local echo signal in a
similar way. That is, 1D Fourier's transform of the searching
region echo signal r'.sub.1(l) [0.ltoreq.l.ltoreq.2L-1] centered on
the point x in the pre-deformation echo signal region is multiplied
with 56 exp { - j 2 L d x i - l ( x ) x l } , ( 58 " )
[0402] or 1D Fourier's transform of the i-1 th phase-matched
searching region echo signal r'.sup.i-1.sub.1 (l) is multiplied
with 57 exp { - j 2 L u ^ x i - l ( x ) x l } . ( 58 '" )
[0403] (Process 2: Estimation of 1D Residual Displacement Component
at the Point x)
[0404] Local 1D echo cross-spectrum is evaluated from the 1D
Fourier's transforms of the pre-deformation local 1D ultrasound
echo signal r.sub.1(l) and phase-matched post-deformation local 1D
ultrasound echo signal r.sup.i.sub.2 (l).
S.sub.2,1.sup.i(l)=R.sub.2.sup.i*(l)R.sub.1(l), (59)
[0405] where * denotes conjugate.
[0406] Alternatively, when pre-deformation local 1D ultrasound echo
signal is phase-matched, cross-spectrum of r.sup.i.sub.1(l) and
r.sub.2(l) is evaluated as
S.sup.i.sub.2,1(l)=R.sub.2*(l)R.sup.i.sub.1(l).
[0407] Cross-spectrum is represented as 58 S 2 , 1 i ( l ) R 1 i (
l ) 2 exp { j 2 L u x i ( x ) x l } , ( 60 )
[0408] where 0.ltoreq.l.ltoreq.L-1,
[0409] and then the phase is represented as 59 i ( l ) = tan - 1 (
Im [ S 2 , 1 i ( l ) ] Re [ S 2 , 1 i ( l ) ] ) , ( 61 )
[0410] where Re[.] and Im[.] respectively represent the real and
imaginary component of ".".
[0411] The least squares method is implemented on the gradient of
the phase eq. (61) weighted with squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertline..sup.2(=Re.sup.2[S.sub.2,1.sup.i(l)-
].sup.2+Im.sup.2[S.sub.2,1.sup.i(l)]). That is, by minimizing
functional: 60 error ( u x i ( x ) ) = l S 2 , 1 i ( l ) 2 ( i ( l
) - u x i ( x ) ( 2 L x ) l ) 2 ( 62 )
[0412] with respect to the 1D residual component u.sub.x.sup.i(x)
to be used to update the i-1 th estimate d x.sup.i-1(x) of the 1D
displacement component dx(x), the estimate of u.sub.x.sup.i(x) is
obtained as .sub.x.sup.i(x) Concretely, the next equation is
solved. 61 l S 2 , 1 i ( l ) 2 ( 2 L x ) l i ( l ) = l S 2 , 1 i (
l ) 2 ( 2 L x ) 2 l 2 u x i ( x ) ( 63 )
[0413] When the 1D displacement component dx(x) is large, the 1D
residual displacement component u.sub.x.sup.i(x) needs to be
estimated after unwrapping the phase of the cross-spectrum [eq.
(59)] in the frequency domain 1.
[0414] Alternatively, when the 1D displacement component dx (x) is
large, by using cross-correlation method (evaluation of the peak
position of the cross-correlation function obtained as 1D inverse
Fourier's transform of the cross-spectrum [eq. (59)]) at the
initial stages during iterative estimation, the 1D residual
displacement component u.sub.x.sup.i(x) can be estimated without
unwrapping the phase of the cross-spectrum [eq. (59)] in the
frequency domain. Specifically, by using the cross-correlation
method, x component of the 1D displacement component is estimated
as integer multiplication of the ultrasound echo sampling interval
.DELTA.x. For instance, with respect to threshold values
correTratio or correTdiff, after 62 ; u ^ x i ( x ) r; ; u ^ x i -
1 ( x ) r; correTratio ( 64 )
[0415] or
.parallel..sub.x.sup.i(x).ltoreq.correTdiff (64')
[0416] is satisfied with where .parallel..sub.x.sup.i(x).parallel.
and .parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components, by using the estimate of the 1D displacement component
dx(x) as the initial estimate, the 1D residual displacement
component is estimated from the gradient of the phase of the
cross-spectrum [eq. (59)].
[0417] Empirically it is known that after using cross-correlation
method the condition
.vertline.u.sup.i.sub.x(x).vertline..ltoreq..DELTA.x/2 is satisfied
with. Then, the necessary and sufficient condition for allowing
estimation of the 1D residual displacement component without
unwrapping the phase of the cross-spectrum 63 u x i ( x ) x 1 ( 65
)
[0418] is satisfied with.
[0419] Alternatively, when the magnitude of the 1D displacement
component dx(x) is large, at initial stages, the acquired original
ultrasound echo data can be thinned out with constant interval in
the direction and the reduced echo data can be used such that the
1D residual displacement component can be estimated without
unwrapping the phase of the cross-spectrum [eq.(59)] in the
frequency domain 1. Specifically, the acquired original ultrasound
echo data are thinned out with constant interval in the direction
and the reduced ehco data are used such that the condition (65) or
(65') is satisfied with.
.vertline.u.sub.x.sup.i(x).vertline..ltoreq..DELTA.x/2 (65')
[0420] The iteration number i increasing, i.e., the magnitude of
the 1D residual displacement component u.sup.i.sub.x(x) decreasing,
the ultrasound echo data density is made restored in the direction,
for instance, twice per iteration.
[0421] The interval of the ultrasound echo signal data are
shortened, for instance, when with respect to threshold values
stepTratio or stepTdiff the condition 64 ; u ^ x i ( x ) r; ; u ^ x
i - 1 ( x ) r; stepTratio ( 66 )
[0422] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..ltoreq.stepTdiff
(66')
[0423] is satisfied with, where .parallel..sub.x.sup.i(x).parallel.
and .parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0424] (Process 3: Update of the 1D Displacement Component Estimate
of the Point x)
[0425] Thus, the i th estimate of the 1D displacement component
dx(x) is evaluated as
dx.sup.i(x)=dx.sup.i-1(x)+.sub.x.sup.i(x). (67)
[0426] [Process 4: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement
(Condition for Making the Local Region Small)]
[0427] In order to make the spatial resolution high of the 1D
displacement component distribution measurement, the local region
is made small during iterative estimation. The criteria is
below-described. The processes 1, 2 and 3 are iteratively carried
out till the criteria is satisfied with. When the criteria is
satisfied with, the local region is made small, for instance, the
length of the local region is made half. For instance, the criteria
is (68) or (68') with respect to threshold values Tratio or Tdiff.
65 ; u ^ x i ( x ) r; ; u ^ x i - 1 ( x ) r; Tratio ( 68 )
[0428] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..ltoreq.Tdiff,
(68')
[0429] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0430] (Process 5: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component of the Point x)
[0431] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component of each
point. The processes 1, 2 and 3 are iteratively carried out till
the criteria is satisfied with. For instance, the criteria is (69)
or (69') with respect to threshold values aboveTratio or
aboveTdiff. 66 ; u ^ x i ( x ) r; ; u ^ x i - 1 ( x ) r;
aboveTratio ( 69 )
[0432] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..ltoreq.aboveTdiff,
(69')
[0433] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0434] (Process 6)
[0435] The 1D displacement component distribution is obtained by
carrying out processes 1, 2, 3, 4, and 5 at every point in the 1D
ROI.
[0436] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0437] [Limitation of Method 3-1]
[0438] The estimate of the 1D displacement component dx(x) is
iteratively updated at each point x in the 1D ROI. Being dependent
on the SNR of the local 1D echo signals, particularly at initial
stages errors possibly occur when estimating the residual component
and then phase matching possibly diverges. For instance, when
solving eq. (63) [process 2] or detecting the peak position of the
cross-correlation function [process 2], errors possibly occur.
[0439] The possibility for divergence of the phase matching is, for
instance, confirmed by the condition (70) or (70') with respect to
the threshold value belowTratio or BelowTdiff. 67 ; u ^ x i ( x )
r; ; u ^ x i - 1 ( x ) r; belowTratio ( 70 )
[0440] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..gtoreq.belowTdiff,
(70')
[0441] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0442] To prevent phase matching (process 1) from diverging, in the
below-described methods 3-2, 3-3, 3-4, and 3-5, by freely using the
condition (70) or (70'), estimation error is reduced of the
residual component. Thus, even if the SNR of the ultrasound echo
signals are low, accurate 1D displacement component measurement can
be realized.
[0443] [Method 3-2]
[0444] The flowchart of the method 3-2 is shown in FIG. 11. To
prevent phase matching from diverging at the process 1 of the
method 3-1, estimation error is reduced of the residual component.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0445] The procedure of iterative estimation is different from that
of the method 3-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0446] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution)
[0447] Phase matching and estimation of the 1D residual
displacement component are performed at every point x in the 1D
ROI. That is, the processes 1 and 2 of the method 3-1 are performed
once at every point in the ROI. Thus, the estimate of the 1D
residual component distribution is obtained.
[0448] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution)
[0449] The i-1 th estimate of the 1D displacement component
distribution is updated using i th estimate of the 1D residual
component distribution.
dx.sup.i(x)={haeck over (d)}.sub.x.sup.i-1(x)+.sub.x.sup.i(x)
(71)
[0450] Next, this estimate is 1D low pass filtered or 1D median
filter to yield the estimate of the 1D displacement component
distribution:
{haeck over (d)}.sub.x.sup.i(x)=LPF[dx.sup.i(x)], or {haeck over
(d)}.sub.x.sup.i(x)=MED[dx.sup.i(x)]. (72)
[0451] Thus, the estimation error is reduced of the residual
component compared with process 2 of the method 3-1 [eq. (63)].
Hence, phase matching of the process 1 of method 3-2 is performed
using smoothed estimate of the 1D displacement component
distribution.
[0452] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement
(Condition for Making the Local Region Small)]
[0453] In order to make the spatial resolution high of the 1D
displacement component distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the ROI is made small.
[0454] The criteria for each point is below-described. The
processes 1 and 2 (method 3-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of the
local region is made half. For instance, the criteria is (73) or
(73') with respect to threshold values Tratio or Tdiff. 68 ; u ^ x
i ( x ) r; ; u ^ x i - 1 ( x ) r; Tratio ( 73 )
[0455] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..ltoreq.Tdiff,
(73')
[0456] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0457] The criteria over the ROI is below-described. The processes
1 and 2 (method 3-2) are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
region is made small, for instance, the length of the local region
is made half. For instance, the criteria is (74) or (74') with
respect to threshold values Tratioroi or Tdiffroi. 69 x ROI ; u ^ x
i ( x ) r; 2 x ROI ; u ^ x i - 1 ( x ) r; 2 Tratioroi or ( 74 ) x
ROI ; u ^ x i ( x ) - u ^ x i - 1 ( x ) r; Tdiffroi , ( 74 ' )
[0458] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0459] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution)
[0460] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution.
The processes 1, 2 and 3 of method 3-2 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(75) or (75') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 70 x ROI ; u ^ x i ( x ) r; 2 x ROI ; u ^ x i - 1 (
x ) r; 2 aboveTratioroi or ( 75 ) x ROI ; u ^ x i ( x ) - u ^ x i -
1 ( x ) r; aboveTdiffroi , ( 75 ' )
[0461] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0462] Final estimate is obtained from eq. (71) or eq. (72).
[0463] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0464] [Method 3-3]
[0465] The flowchart of the method 3-3 is shown in FIG. 12. To
prevent phase matching from diverging at the process 1 of the
method 3-1, estimation error is reduced of the residual component.
Possibility of divergence is detected from above-described
condition (70) or (70'), and by effectively utilizing method 3-1
and 3-2, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0466] At first, the procedure of iterative estimation is same as
that of the method 3-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0467] Phase matching and estimation of the 1D residual
displacement component are performed at every point x in the 1D
ROI. That is, the processes 1 and 2 of the method 3-1 are performed
once at every point in the ROI. Thus, the estimate of the 1D
residual component distribution is obtained.
[0468] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 3-1 is used. If condition (70) or
(70') is satisfied with at points or regions, in the process 2 of
the method 3-2, over sufficiently large regions centered on the
points or regions, or over the ROI, the estimate dx.sup.i(x) of the
1D displacement component dx(x) can be 1D low pass filtered or 1D
median filtered as eq. (76).
{haeck over (d)}.sub.x.sup.i(x)=LPF[dx.sup.i(x)], or {haeck over
(d)}.sub.x.sup.i(x)=MED[dx.sup.i(x)] (76)
[0469] Thus, the estimation error is reduced of the residual
component compared with process 2 of the method 3-1 [eq. (63)].
[0470] Thus, iterative estimation is terminated at the process 5 of
the method 3-1 or the process 4 of the method 3-2. Hence, final
estimate is obtained from eq. (67), or eq.(71), or eq.(76).
[0471] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0472] [Method 3-4]
[0473] The flowchart of the method 3-4 is shown in FIG. 13. To
prevent phase matching from diverging at the process 1 of the
method 3-1, estimation error is reduced of the residual component.
Thus, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0474] The procedure of iterative estimation is different from that
of the method 3-1. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0475] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution)
[0476] Phase matching and estimation of the 1D residual
displacement component are performed at every point x in the 1D
ROI. That is, the process 1 of the method 3-1 is performed once at
every point in the ROI.
[0477] To obtain the estimate .sub.x.sup.i(x) of the residual
component distribution u.sup.i.sub.x(x), at every point local 1D
echo cross-spectrum is evaluated from the 1D Fourier's transforms
of the pre-deformation local 1D ultrasound echo signal r.sub.1(l)
and phase-matched post-deformation local 1D ultrasound echo signal
r.sup.i.sub.2 (l). Alternatively, when pre-deformation local 1D
ultrasound echo signal is phase-matched, at every point
cross-spectrum of r.sup.i.sub.1(l) and r.sub.2(l) is evaluated.
[0478] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertline..sup.2, where each weight
function is normalized by the power of the cross-spectrum, i.e., 71
l | S 2 , 1 i ( l ) | 2 .
[0479] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 1D residual component distribution
u.sub.x.sup.i(x).
error
(u.sup.i)=.vertline..vertline.a-Fu.sup.i.vertline..vertline..sup.2+.-
alpha..sub.1i.vertline..vertline.u.sup.i.vertline..vertline..sup.2+.alpha.-
.sub.2i.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2+.alpha..sub.-
3i.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2+.alpha..su-
b.4i.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2
(77)
[0480] where
[0481] a: vector comprised of x distribution of the cross-spectrum
phase .THETA..sup.i(1) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l- ).vertline. normalized by the
magnitude of the cross-spectrum 72 l | s 2 , 1 i ( l ) | 2
[0482] evaluated at every point in the 1D ROI.
[0483] F: matrix comprised of x distribution of the Fourier's
coordinate value l weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertlin- e. normalized by the
magnitude of the cross-spectrum 73 l | s 2 , 1 i ( l ) | 2
[0484] evaluated at every point in the 1D ROI.
[0485] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0486] Gu.sup.i: vector comprised of the finite difference
approximations of the 1D distribution of the 1D gradient components
of the unknown 1D residual components u.sub.x.sup.i(x) 74 x 2 u x i
( x )
[0487] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 1D distribution of the 1D Laplacians of the
unknown 1D residual components u.sub.x.sup.i(x) 75 2 x 2 u x i ( x
)
[0488] GG.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 1D distribution of the 1D gradient components
of the 1D Laplacians of the unknown 1D residual components
u.sub.x.sup.i(x) 76 x ( 2 x 2 u x i ( x ) )
[0489] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error(u.sup.i) has one minimum value. Thus, by
solving for residual displacement component distribution
u.sub.x.sup.i(x) the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(78)
[0490] estimate .sub.x.sup.i(x) of the residual component
distribution u.sup.i.sub.x(x) is stably obtained. Thus, estimation
error is reduced of the residual component.
[0491] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement component etc.
[0492] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution)
[0493] The i-1 th estimate of the 1D displacement component
distribution is updated using i th estimate of the 1D residual
component distribution.
dx.sup.i(x)={haeck over (d)}.sub.x.sup.i-1(x)+.sub.x.sup.i(x)
(79)
[0494] Freely, this estimate can be 1D low pass filtered or 1D
median filter to yield the estimate of the 1D displacement
component distribution.
{haeck over (d)}.sub.x.sup.i(x)=LPF[dx.sup.i(x)], or {haeck over
(d)}.sub.x.sup.i(x)=MED[dx.sup.i(x)] (80)
[0495] Hence, phase matching of the process 1 of method 3-4 is
performed using the 1D residual component data u.sub.x.sup.i(x)
obtained from eq. (78), or the 1D component data dx.sup.i(x)
obtained from eq. (79), or smoothed estimate obtained from eq.
(80).
[0496] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement
(Condition for Making the Local Region Small)]
[0497] In order to make the spatial resolution high of the 1D
displacement component distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the ROI is made small.
[0498] The criteria for each point is below-described. The
processes 1 and 2 of method 3-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
the local region is made half. For instance, the criteria is (81)
or (81') with respect to threshold values Tratio or Tdiff. 77 u ^ x
i ( x ) u ^ x i - 1 ( x ) Tratio ( 81 )
[0499] or
.parallel..sub.x.sup.i(x)-.sub.x.sup.i-1(x).parallel..ltoreq.Tdiff,
(81')
[0500] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0501] The criteria over the ROI is below-described. The processes
1 and 2 of method 3-4 are iteratively carried out till the criteria
is satisfied with. When the criteria is satisfied with, the local
region is made small, for instance, the length of the local region
is made half. For instance, the criteria is (82) or (82') with
respect to threshold values Tratioroi or Tdiffroi. 78 xROI u ^ x i
( x ) 2 xROI u ^ x i - 1 ( x ) 2 Tratioroi ( 82 ) or xROI u ^ x i (
x ) - u ^ x i - 1 ( x ) Tdiffroi , ( 82 ' )
[0502] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0503] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution)
[0504] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution.
The processes 1, 2 and 3 of method 3-4 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(83) or (83') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 79 xROI u ^ x i ( x ) | 2 xROI u ^ x i - 1 ( x ) 2
aboveTratioroi ( 83 ) xROI u ^ x i ( x ) - u ^ x i - 1 ( x ) above
Tdiffroi, ( 83 ' )
[0505] where .parallel..sub.x.sup.i(x).parallel. and
.parallel..sub.x.sup.i-1(x).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0506] Final estimate is obtained from eq. (79) or eq. (80).
[0507] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0508] [Method 3-5]
[0509] The flowchart of the method 3-5 is shown in FIG. 14. To
prevent phase matching from diverging at the process 1 of the
method 3-1, estimation error is reduced of the residual component.
Possibility of divergence is detected from above-described
condition (70) or (70'), and by effectively utilizing method 3-1
and 3-4, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0510] At first, the procedure of iterative estimation is same as
that of the method 3-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0511] Phase matching and estimation of the 1D residual
displacement component are performed at every point x in the 1D
ROI. That is, the process 1 of the method 3-1 is performed once at
every point in the ROI. Moreover, using the regularization method,
stably the estimate of the 1D residual component distribution is
obtained.
[0512] i-1 th estimate {haeck over (d)}.sub.x.sup.i-1(x) of 1D
displacement component distribution dx(x).
[0513] i th estimate .sub.x.sup.i(x) of 1D residual component
distribution u.sub.x.sup.i(x)
[0514] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 3-1 is used. If condition (70) or
(70') is satisfied with at points or regions, in the process 2 of
the method 3-4, over sufficiently large regions centered on the
points or regions, or over the ROI, the estimate dx.sup.i(x) of the
1D displacement component dx(x) can be 1D low pass filtered or 1D
median filtered as eq. (84).
{haeck over (d)}.sub.x.sup.i(x)=LPF[dx.sup.i(x)], or {haeck over
(d)}.sub.x.sup.i(x)=MED[dx.sup.i(x)] (84)
[0515] Thus, the estimation error is reduced of the residual
component.
[0516] Iterative estimation is terminated at the process 5 of the
method 3-1 or the process 4 of the method 3-4. Hence, final
estimate is obtained from eq. (67), or eq.(79), or eq.(84).
[0517] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0518] (IV) Method 4: Measurement of 2D Displacement Component
Vector Distribution in 3D SOI
[0519] [Method 4-1]
[0520] 2D displacement component vector distribution in 3D SOI can
be measured by measuring 2D displacement component vector
distribution in each (x,y) plane by means of the method 2-1, or
2-2, or 2-3, or 2-4, or 2-5 (FIG. 21).
[0521] (Process 1)
[0522] In each (x,y) plane in 3D SOI, the method 2-1, or 2-2, or
2-3, or 2-4, or 2-5 is utilized. The initial estimate [eq. (29)] of
the iterative estimation of the 2D displacement vector in the 3D
SOI is set as zero vector if a priori data is not given about
displacements of body motion nor applied compression.
Alternatively, values accurately estimated at neighborhood can be
used (high correlation or least squares).
[0523] Moreover, the methods 4-2, 4-3, 4-4, and 4-5 are
respectively baesd on the methods 2-2, 2-3, 2-4, and 2-5.
[0524] [Method 4-2]
[0525] The flowchart of the method 4-2 is shown in FIG. 22. As
example, let's consider measurement of 2D displacement vector
d(x,y,z) [=(dx(x,y,z),dy(x,y,z)).sup.T] in 3D SOI. At i th estimate
(i.gtoreq.1), the following processes are performed.
[0526] (Process 1: Estimation of the 2D Residual Displacement
Component Vector Distribution in 3D SOI)
[0527] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the processes 1 and 2 of the method 2-1 are performed
once at every point in the 3D SOI. Thus, the i th estimate of the
2D residual component vector distribution u.sup.i(x,y,z) is
obtained as
.sup.i(x,y,z)[=(.sub.x.sup.i(x,y,z),
.sub.y.sup.i(x,y,z)).sup.T].
[0528] (Process 2: Update of the Estimate of the 2D Displacement
Component Vector Distribution in 3D SOI)
[0529] The i-1 th estimate of the 2D displacement component vector
distribution in the 3D SOI is updated using i th estimate of the 2D
residual component vector distribution in the 3D SOI.
d.sup.i(x,y,z)={haeck over (d)}.sup.i-1(x,y,z)+.sub.i(x,y,z)
(85)
[0530] Next, this estimate is 3D low pass filtered or 3D median
filter to yield the estimate of the 2D displacement component
vector distribution:
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)]. (86)
[0531] Thus, the estimation error is reduced of the residual vector
compared with process 2 of the method 2-1 [eq. (35)]. Hence, phase
matching of the process 1 of method 4-2 is performed using smoothed
estimate {haeck over (d)}.sup.i(x,y,z) of the 2D displacement
component vector distribution in the 3D SOI.
[0532] [Process 3: Condition for Heightening the Spatial Resolution
of the 2D Displacement Component Vector Distribution Measurement in
3D SOI (Condition for Making the Local Region Small)]
[0533] In order to make the spatial resolution high of the 2D
displacement component vector distribution measurement in the 3D
SOI, during iterative estimation, the local region used for each
point is made small, or the local region used over the 3D SOI is
made small.
[0534] The criteria for each point is below-described. The
processes 1 and 2 (method 4-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of each
side is made half. For instance, the criteria is (87) or (87') with
respect to threshold values Tratio or Tdiff. 80 u ^ i ( x , y , z )
u ^ i - 1 ( x , y , z ) Tratio ( 87 )
[0535] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.Tdiff,
(87')
[0536] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0537] The condition (87) or (87') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0538] The criteria over the 3D SOI is below-described. The
processes 1 and 2 (method 4-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of each
side is made half. For instance, the criteria is (88) or (88') with
respect to threshold values Tratioroi or Tdiffroi. 81 ( x , y , z )
SOI ; u i ( x , y , z ) r; 2 ( x , y , z ) SOI ; u i - 1 ( x , y ,
z ) r; 2 Tratioroi or ( 88 ) ( x , y , z ) SOI ; u i ( x , y , z )
- u i - 1 ( x , y , z ) r; Tdiffroi , ( 88 ' )
[0539] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0540] The condition (88) or (88') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0541] (Process 4: Condition for Terminating the Iterative
Estimation of the 2D Displacement Component Vector Distribution in
3D SOI)
[0542] Below-described is the criteria for terminating the
iterative estimation of the 2D displacement component vector
distribution in the 3D SOI. The processes 1, 2 and 3 of method 4-2
are iteratively carried out till the criteria is satisfied with.
For instance, the criteria is (89) or (89') with respect to
threshold values aboveTratioroi or aboveTdiffroi. 82 ( x , y , z )
SOI ; u i ( x , y , z ) r; 2 ( x , y , z ) SOI ; u i - 1 ( x , y ,
z ) r; 2 aboveTratioroi or ( 89 ) ( x , y , z ) SOI ; u i ( x , y ,
z ) - u i - 1 ( x , y , z ) r; aboveTdiffroi , ( 89 ' )
[0543] where .parallel..sup.i(x,y,z).parallel. and
.sup.i-1(x,y,z).paralle- l. are respectively norms (magnitudes) of
the i th and i-1 th estimates of the residual vectors.
[0544] Final estimate is obtained from eq. (85) or eq. (86).
[0545] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0546] [Method 4-3]
[0547] The flowchart of the method 4-3 is shown in FIG. 23. As
example, let's consider measurement of 2D displacement vector
d(x,y,z)[=(dx(x,y,z),dy(x,y,z)).sup.T] in 3D SOI.
[0548] Possibility of divergence is detected from above-described
condition (42) or (42') in above-described process 1 of the method
4-2, and by effectively utilizing the method 4-1 based on the
method 2-1, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0549] At first, the procedure of iterative estimation is same as
that of the method 4-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0550] The process 1 of the method 4-2 is performed. (Phase
matching and estimation of the 2D residual displacement vector are
performed at every point (x,y,z) in the 3D SOI.) That is, the
processes 1 and 2 of the method 2-1 are performed once at every
point in the 3D SOI. Thus, the estimate of the 2D residual
component vector distribution is obtained.
[0551] During this estimation, if neither condition (42) nor (42')
is satisfied with, the method 4-1 is used. If condition (42) or
(42') is satisfied with at points or regions, in the process 2 of
the method 4-2, over sufficiently large regions centered on the
points or regions, or over the 3D SOI, the estimated (x,y,z) [eq.
(85)] of the 2D displacement vector d (x,y,z) can be 3D low pass
filtered or 3D median filtered as eq. (90). Thus, the estimation
error is reduced of the residual vector compared with process 2 of
the method 2-1 [eq.(35)].
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (90)
[0552] Thus, iterative estimation is terminated at the process 1 of
the method 4-1 based on the method 2-1, or the process 4 of the
method 4-2. Hence, final estimate is obtained from eq. (39), or eq.
(85), or eq. (90).
[0553] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0554] [Method 4-4]
[0555] The flowchart of the method 4-4 is shown in FIG. 24. As
example, let's consider measurement of 2D displacement vector
d(x,y,z)[=(dx(x,y,z),dy(x,y,z)).sup.T] in 3D SOI. At i th estimate
(i.gtoreq.1), the following process 1 is performed.
[0556] (Process 1: Estimation of the 2D Residual Displacement
Component Vector Distribution in 3D SOI)
[0557] Phase matching and estimation of the 2D residual
displacement vector are performed at every point (x,y,z) in the 3D
SOI. That is, the process 1 of the method 2-1 is performed once at
every point in the 3D SOI.
[0558] To obtain the estimate .sup.i(x,y,z)[=.sub.x.sup.i(x,y,z),
.sub.y.sup.i(x,y,z)).sup.T] of the 2D residual component vector
distribution u.sup.i(x,y,z)[=(u.sup.i.sub.x(x,y,z),
u.sup.i.sub.y(x,y,z)).sup.T] in the 3D SOI, at every point local 2D
echo cross-spectrum [eq. (31)] is evaluated from the 2D Fourier's
transforms of the pre-deformation local 2D ultrasound echo signal
r.sub.1(l,m) and phase-matched post-deformation local 2D ultrasound
echo signal r.sup.i.sub.2 (l,m) Alternatively, when pre-deformation
local 2D ultrasound echo signal is phase-matched, at every point
cross-spectrum of r.sup.i.sub.1(l,m) and r.sub.2(l,m) is
evaluated.
[0559] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum .vertline.S.sub.2,1.sup.i(l,m).vertline.,
where each weight function is normalized by the power of the
cross-spectrum, 83 i . e . , l , m S 2 , 1 i ( l , m ) 2 . [ eq . (
34 ) ]
[0560] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 2D residual component vector distribution
u.sup.i(x,y,z) in the 3D SOI.
error
(u.sup.i)=.vertline..vertline.a-Fu.sup.i.vertline..vertline..sup.2+.-
alpha..sub.1i.vertline..vertline.u.sup.i.vertline..vertline..sup.2+.alpha.-
.sub.2i.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2+.alpha..sub.-
3i.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2+.alpha..su-
b.4i.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2
(91)
[0561] where
[0562] a: vector comprised of (x,y,z) distribution of the
cross-spectrum phase .THETA..sup.i(l,m) weighted with
cross-spectrum .vertline.S.sub.2,1.sup.i(l,m).vertline. normalized
by the magnitude of the cross-spectrum 84 l , m S 2 , 1 i ( l , m )
2
[0563] evaluated at every point in the 3D SOI.
[0564] F: matrix comprised of (x,y,z) distribution of the Fourier's
coordinate value (l,m) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l,m).vertline. normalized by the
magnitude of the cross-spectrum 85 l , m S 2 , 1 i ( l , m ) 2
[0565] evaluated at every point in the 3D SOI.
[0566] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0567] Gu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the unknown 2D residual vector u.sup.i(x,y,z)
components 86 x u x i ( x , y , z ) , y u x i ( x , y , z ) , z u x
i ( x , y , z ) , x u y i ( x , y , z ) , y u y i ( x , y , z ) , z
u y i ( x , y , z )
[0568] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D Laplacian of the
unknown 2D residual vector u.sup.i(x,y,z) components 87 2 x 2 u x i
( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z
) 2 x 2 u y i ( x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z 2 u y
i ( x , y , z )
[0569] GG.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the 3D Laplacians of the unknown 2D residual vector
u.sup.i(x,y,z) components 88 x ( 2 x 2 u x i ( x , y , z ) + 2 y 2
u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z ) ) , y ( 2 x 2 u x i
( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z
) ) , z ( 2 x 2 u x i ( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2
z 2 u x i ( x , y , z ) ) , x ( 2 x 2 u y i ( x , y , z ) + 2 y 2 u
y i ( x , y , z ) + 2 z 2 u y i ( x , y , z ) ) , y ( 2 x 2 u y i (
x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z 2 u y i ( x , y , z )
) , z ( 2 x 2 u y i ( x , y , z ) + 2 y 2 u y i ( x , y , z ) + 2 z
2 u y i ( x , y , z ) )
[0570] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error (u.sup.i) has one minimum value. Thus, by
solving for 2D residual displacement component vector distribution
u.sup.i(x,y,z) in the 3D SOI the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(92-1)
[0571] estimate .sup.i(x,y,z)[=(.sub.x.sup.i(x,y,z),
.sub.y.sup.i(x,y,z)).sup.T] of the 2D residual component vector
distribution u.sup.i(x,y,z)[=(u.sup.i.sub.x(x,y,z),
u.sup.i.sub.y(x,y,z)).sup.T] is stably obtained. Thus, estimation
error is reduced of the residual vector.
[0572] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement vector, direction of the unknown displacement
component, direction of the partial derivative, etc.
[0573] (Process 2: Update of the Estimate of the 2D Displacement
Component Vector Distribution)
[0574] The i-1 th estimate of the 2D displacement component vector
distribution is updated using i th estimate of the 2D residual
component vector distribution.
d.sup.i(x,y,z)={haeck over (d)}.sup.i-1(x,y,z)+.sup.i(x,y,z)
(92-2)
[0575] Freely, this estimate can be 3D low pass filtered or 3D
median filter to yield the estimate of the 2D displacement
component vector distribution.
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (93)
[0576] Hence, phase matching of the process 1 of method 4-4 is
performed using the 2D residual vector data u.sup.i(x,y,z) obtained
from eq. (91), or the 2D vector data d.sup.i(x,y,z) obtained from
eq. (92-2), or smoothed estimate obtained from eq. (93).
[0577] [Process 3: Condition for Heightening the Spatial Resolution
of the 2D Displacement Component Vector Distribution Measurement in
3D SOI (Condition for Making the Local Region Small)]
[0578] In order to make the spatial resolution high of the 2D
displacement component vector distribution measurement, during
iterative estimation, the local region used for each point is made
small, or the local region used over the 3D SOI is made small.
[0579] The criteria for each point is below-described. The
processes 1 and 2 of method 4-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
each side is made half. For instance, the criteria is (94) or (94')
with respect to threshold values Tratio or Tdiff. 89 ; u i ( x , y
, z ) r; ; u i - 1 ( x , y , z ) r; Tratio ( 94 )
[0580] or
.parallel..sup.i(x,y,z)-.sup.i-1(x,y,z).parallel..ltoreq.Tdiff,
(94')
[0581] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0582] The condition (94) or (94') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0583] The criteria over the 3D SOI is below-described. The
processes 1 and 2 of method 4-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
each side is made half. For instance, the criteria is (95) or (95')
with respect to threshold values Tratioroi or Tdiffroi. 90 ( x , y
, z ) SOI ; u i ( x , y , z ) r; 2 ( x , y , z ) SOI ; u i - 1 ( x
, y , z ) r; 2 Tratioroi or ( 95 ) ( x , y , z ) SOI ; u i ( x , y
, z ) - u i - 1 ( x , y , z ) r; Tdiffroi , ( 95 ' )
[0584] where .parallel..sup.i(x,y,z).parallel. and
.parallel..sup.i-1(x,y,- z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
vectors.
[0585] The condition (95) or (95') can be applied to each direction
component, and in this case the side is shorten in each
direction.
[0586] (Process 4: Condition for Terminating the Iterative
Estimation of the 2D Displacement Component Vector Distribution in
3D SOI)
[0587] Below-described is the criteria for terminating the
iterative estimation of the 2D displacement component vector
distribution. The processes 1, 2 and 3 of method 4-4 are
iteratively carried out till the criteria is satisfied with. For
instance, the criteria is (96) or (96') with respect to threshold
values aboveTratioroi or aboveTdiffro. 91 ( x , y , z ) S0I ; i ( x
, y , z ) r; 2 ( x , y , z ) S0I ; i - 1 ( x , y , z ) r; 2
aboveTratioroi or ( 96 ) ( x , y , z ) S0I ; i ( x , y , z ) - i -
1 ( x , y , z ) r; above Tdiffroi , ( 96 ' )
[0588] where .parallel..sup.i(x,y,z).parallel. and
.sup.i-1(x,y,z).paralle- l. are respectively norms (magnitudes) of
the i th and i-1 th estimates of the residual vectors.
[0589] Final estimate is obtained from eq. (92-2) or eq. (93).
[0590] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0591] [Method 4-5]
[0592] The flowchart of the method 4-5 is shown in FIG. 25. As
example, let's consider measurement of 2D displacement vector
d(x,y,z)[=(dx(x,y,z),dy(x,y,z)).sup.T] in 3D SOI.
[0593] Possibility of divergence is detected from above-described
condition (42) or (42') in above-described process 1 of the method
4-4, and by effectively utilizing the method 4-1 based on the
method 2-1, even if the SNR of the ultrasound echo signals are low,
accurate 2D displacement vector measurement can be realized.
[0594] At first, the procedure of iterative estimation is same as
that of the method 4-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0595] The process 1 of the method 4-4 is performed. (Phase
matching and estimation of the 2D residual displacement vector are
performed at every point (x,y,z) in the 3D SOI.) That is, the
process 1 of the method 2-1 is performed once at every point in the
3D SOI. Moreover, using the regularization method, stably the
estimate of the 2D residual component vector distribution is
obtained.
[0596] i-1 th estimate {haeck over (d)}.sup.i-1(x,y,z) of 2D
displacement component vector distribution d(x,y,z).
[0597] i th estimate {haeck over (u)}.sup.i(x,y,z) of 2D residual
component vector distribution u.sup.i(x,y,z).
[0598] During this estimation, if neither condition (42) nor (42')
is satisfied with, the method 4-1 based on the method 2-1 is used.
If condition (42) or (42') is satisfied with at points or regions,
in the process 2 of the method 4-4, over sufficiently large regions
centered on the points or regions, or over the 3D SOI, the estimate
d.sup.i(x,y,z) of the 2D displacement vectord (x,y,z) can be 3D low
pass filtered or 3D median filtered as eq. (97).
{haeck over (d)}.sup.i(x,y,z)=LPF[d.sup.i(x,y,z)], or {haeck over
(d)}.sup.i(x,y,z)=MED[d.sup.i(x,y,z)] (97)
[0599] Thus, the estimation error is reduced of the residual
vector.
[0600] Iterative estimation is terminated at the process 1 of the
method 4-1 based on the method 2-1, or the process 4 of the method
4-4. Hence, final estimate is obtained from eq. (39), or eq.
(92-2), or eq. (97).
[0601] The initial estimate [eq. (29)] of the iterative estimation
of the 2D displacement vector is set as zero vector if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0602] (V) Method 5: Measurement of 1D Displacement (one Direction)
Component Distribution in 3D SOI
[0603] [Method 5-1]
[0604] 1D x displacement component distribution in 3D SOI can be
measured by measuring 1D x displacement component distribution in
each line being parallel to x axis by means of the method 3-1, or
3-2, or 3-3, or 3-4, or 3-5 (FIG. 21).
[0605] (Process 1)
[0606] In each line being parallel to x axis in 3D SOI, the method
3-1, or 3-2, or 3-3, or 3-4, or 3-5 is utilized. The initial
estimate [eq. (57)] of the iterative estimation of the 1D
displacement component in the 3D SOI is set as zero if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0607] Moreover, the methods 5-2, 5-3, 5-4, and 5-5 are
respectively baesd on the methods 3-2, 3-3, 3-4, and 3-5.
[0608] [Method 5-2]
[0609] The flowchart of the method 5-2 is shown in FIG. 22. As
example, let's consider measurement of 1D displacement component
dx(x,y,z) in 3D SOI. At i th estimate (i.gtoreq.1), the following
processes are performed.
[0610] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution in 3D SOI)
[0611] Phase matching and estimation of the 1D residual
displacement component are performed at every point (x,y,z) in the
3D SOI. That is, the processes 1 and 2 of the method 3-1 are
performed once at every point in the 3D SOI. Thus, the i th
estimate of the 1D residual component distribution
u.sub.x.sup.i(x,y,z) is obtained as .sub.x.sup.i(x,y,z).
[0612] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution in 3D SOI)
[0613] The i-1 th estimate of the 1D displacement component
distribution in the 3D SOI is updated using i th estimate of the 1D
residual component distribution in the 3D SOI.
dx.sup.i(x,y,z)={haeck over
(d)}.sub.x.sup.i-1(x,y,z)+.sub.x.sup.i(x,y,z) (98)
[0614] Next, this estimate is 3D low pass filtered or 3D median
filter to yield the estimate of the 1D displacement component
distribution:
{haeck over (d)}.sub.x.sup.i(x,y,z)=LPF[dx.sup.i(x,y,z)], or {haeck
over (d)}.sub.x.sup.i(x,y,z)=MED[dx.sup.i(x,y,z)]. (99)
[0615] Thus, the estimation error is reduced of the residual
component compared with process 2 of the method3-1 (eq. (63)].
Hence, phase matching of the process 1 of method 5-2 is performed
using smoothed estimate {haeck over (d)}.sub.x.sup.i(x,y,z) of the
1D displacement component distribution in the 3D SOI.
[0616] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement in 3D SOI
(Condition for Making the Local Region Small)]
[0617] In order to make the spatial resolution high of the 1D
displacement component distribution measurement in the 3D SOI,
during iterative estimation, the local region used for each point
is made small, or the local region used over the 3D SOI is made
small.
[0618] The criteria for each point is below-described. The
processes 1 and 2 (method 5-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of the
local region is made half. For instance, the criteriais (87) or
(87') withrespectto thresholdvalues Tratio or Tdiff. 92 ; x i ( x ,
y , z ) r; ; x i - 1 ( x , y , z ) r; Tratio ( 100 )
[0619] or
.parallel..sub.x.sup.i(x,y,z)-.sub.x.sup.i-1(x,y,z).parallel..ltoreq.Tdiff-
, (100')
[0620] where .parallel..sub.x.sup.i(x,y,z).parallel. and
.parallel..sub.x.sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0621] The criteria over the 3D SOI is below-described. The
processes 1 and 2 (method 5-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of the
local region is made half. For instance, the criteria is (101) or
(101') with respect to threshold values Tratioroi or Tdiffroi. 93 (
x , y , z ) S0I ; x i ( x , y , z ) r; 2 ( x , y , z ) S0I ; x i -
1 ( x , y , z ) r; 2 Tratioroi or ( 101 ) ( x , y , z ) S0I ; x i (
x , y , z ) - x i - 1 ( x , y , z ) r; Tdiffroi , ( 101 ' )
[0622] where .parallel..sub.x.sup.i(x,y,z).parallel. and
.parallel..sub.x.sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0623] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution in 3D
SOI)
[0624] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution
in the 3D SOI. The processes 1, 2 and 3 of method 5-2 are
iteratively carried out till the criteria is satisfied with.
Forinstance, thecriteriais (102) or (102') withrespect to threshold
values aboveTratioroi or aboveTdiffroi. 94 ( x , y , z ) S0I ; x i
( x , y , z ) r; 2 ( x , y , z ) S0I ; x i - 1 ( x , y , z ) r; 2
above Tratioroi or ( 102 ) ( x , y , z ) S0I ; x i ( x , y , z ) -
x i - 1 ( x , y , z ) r; above Tdiffroi , ( 102 ' )
[0625] where .parallel..sub.x.sup.i(x,y,z).parallel. and
.parallel..sub.x.sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0626] Final estimate is obtained from eq. (98) or eq. (99).
[0627] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0628] [Method 5-3]
[0629] The flowchart of the method 5-3 is shown in FIG. 23. As
example, let's consider measurement of 1D displacement component
dx(x,y,z) in 3D SOI.
[0630] Possibility of divergence is detected from above-described
condition (70) or (70') in above-described process 1 of the method
5-2, and by effectively utilizing the method 5-1 based on the
method 3-1, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0631] At first, the procedure of iterative estimation is same as
that of the method 5-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0632] The process 1 of the method 5-2 is performed. (Phase
matching and estimation of the 1D residual displacement component
are performed at every point (x,y,z) in the 3D SOI.) That is, the
processes 1 and 2 of the method 3-1 are performed once at every
point in the 3D SOI. Thus, the estimate of the 1D residual
component distribution is obtained.
[0633] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 5-1 is used. If condition (70) or
(70') is satisfied with at points or regions, in the process 2 of
the method 5-2, over sufficiently large regions centered on the
points or regions, or over the 3D SOI, the estimate dx(x,y,z) [eq.
(98)] of the 1D displacement componentdx(x,y,z) can be 3D low pass
filtered or 3D median filtered as eq. (102). Thus, the estimation
error is reduced of the residual component compared with process 2
of the method 3-1 [eq. (63)].
{haeck over (d)}.sub.x.sup.i(x,y,z)=LPF[dx.sup.i(x,y,z)], or {haeck
over (d)}.sub.x.sup.i(x,y,z)=MED[dx.sup.i(x,y,z)] (102)
[0634] Thus, iterative estimation is terminated at the process 1 of
the method 5-1 based on the method 3-1, or the process 4 of the
method 5-2. Hence, final estimate is obtained from eq. (67), or
eq.(98), or eq.(102).
[0635] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0636] [Method 5-4]
[0637] The flowchart of the method 5-4 is shown in FIG. 24. As
example, let's consider measurement of 1D displacement component
dx(x,y,z) in 3D SOI. At i th estimate (i.gtoreq.1), the following
process 1 is performed.
[0638] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution in 3D SOI)
[0639] Phase matching and estimation of the 1D residual
displacement component are performed at every point (x,y,z) in the
3D SOI. That is, the process 1 of the method 3-1 is performed once
at every point in the 3D SOI.
[0640] To obtain the estimate .sub.x.sup.i(x,y,z) of the 1D
residual component distribution u.sup.i.sub.x(x,y,z) in the 3D SOI,
at every point local 1D echo cross-spectrum [eq. (59)] is evaluated
from the 1D Fourier's transforms of the pre-deformation local 1D
ultrasound echo signal r.sub.1(l) and phase-matched
post-deformation local 1D ultrasound echo signal r.sup.i.sub.2 (l)
Alternatively, when pre-deformation local 1D ultrasound echo signal
is phase-matched, at every point cross-spectrum of r.sup.i.sub.1(l)
and r.sub.2(l) is evaluated.
[0641] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum .vertline.S.sub.2,1.sup.i(l).vertline.,
where each weight function is normalized by the power of the
cross-spectrum, i.e., 95 l S 2 , 1 i ( l ) 2 . [ eq . ( 62 ) ]
[0642] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 1D residual component distribution
u.sub.x.sup.i(x,y,z) in the 3D SOI.
error
(u.sup.i)=.vertline..vertline.a-Fu.sup.i.vertline..vertline..sup.2+.-
alpha..sub.1i.vertline..vertline.u.sup.i.vertline..vertline..sup.2+.alpha.-
.sub.2i.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2+.alpha..sub.-
3i.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2+.alpha..su-
b.4i.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2
(103)
[0643] where
[0644] a: vector comprised of (x,y,z) distribution of the
cross-spectrum phase .THETA..sup.i(l) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertline. normalizedby the magnitude
of the cross-spectrum 96 l S 2 , 1 i ( l ) 2
[0645] evaluated at every point in the 3D SOI.
[0646] F: matrix comprised of (x,y,z) distribution of the Fourier's
coordinate value 1 weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i- (l).vertline. normalized by the
magnitude of the cross-spectrum 97 l S 2 , 1 i ( l ) 2
[0647] evaluated at every point in the 3D SOI.
[0648] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0649] Gu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the unknown 1D residual component
u.sub.x.sup.i(x,y,z) 98 x u x i ( x , y , z ) , y u x i ( x , y , z
) , z u x i ( x , y , z )
[0650] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 3D distributions of the 3D Laplacian of the
unknown 1D residual component u.sup.i(x,y,z) 99 2 x 2 u x i ( x , y
, z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z )
[0651] GG.sup.TGu.sup.u: vector comprised of the finite difference
approximations of the 3D distributions of the 3D gradient
components of the 3D Laplacians of the unknown 1D residual
component u.sub.x.sup.i(x,y,z) 100 x ( 2 x 2 u x i ( x , y , z ) +
2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x , y , z ) ) , y ( 2 x 2
u x i ( x , y , z ) + 2 y 2 u x i ( x , y , z ) + 2 z 2 u x i ( x ,
y , z ) ) , z ( 2 x 2 u x i ( x , y , z ) + 2 y 2 u x i ( x , y , z
) + 2 z 2 u x i ( x , y , z ) )
[0652] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error(u.sup.i) has one minimum value. Thus, by
solving for 1D residual displacement component distribution
u.sub.x.sup.i(x,y,z) in the 3D SOI the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(104)
[0653] estimate .sub.x.sup.i(x,y,z) of the 1D residual component
distribution u.sup.i.sub.x(x,y,z) is stably obtained. Thus,
estimation error is reduced of the residual component.
[0654] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement component etc.
[0655] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution)
[0656] The i-1 th estimate of the 1D displacement component
distribution is updated using i th estimate of the 1D residual
component distribution.
dx.sup.i(x,y,z)={haeck over
(d)}.sub.x.sup.i-1(x,y,z)+.sub.x.sup.i(x,y,z) (105)
[0657] Freely, this estimate can be 3D low pass filtered or 3D
median filter to yield the estimate of the 1D displacement
component distribution.
{haeck over (d)}.sup.i(x,y,z)=LPF[dx.sup.i(x,y,z)], or {haeck over
(d)}.sub.x.sup.i(x,y,z)=MED[dx.sup.i(x,y,z)] (106)
[0658] Hence, phase matching of the process 1 of method 5-4 is
performed using the 1D residual component data u.sub.x.sup.i(x,y,z)
obtained from eq. (104), or the 1D component data
d.sub.x.sup.i(x,y,z) obtained from eq. (105), or smoothed estimate
obtained from eq. (106).
[0659] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement in 3D SOI
(Condition for Making the Local Region Small)]
[0660] In order to make the spatial resolution high of the 1D
displacement component distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the 3D SOI is made small.
[0661] The criteria for each point is below-described. The
processes 1 and 2 of method 5-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
the local region is made half. For instance, the criteria is (107)
or (107') with respect to threshold values Tratio or Tdiff. 101 ; u
x i ( x , y , z ) r; ; u x i - 1 ( x , y , z ) r; Tratio ( 107
)
[0662] or
.parallel..sub.x.sup.i(x,y,z)-.sub.x.sup.i-1(x,y,z).ltoreq.Tdiff,
(107')
[0663] where .parallel..sub.x.sup.i(x,y,z).parallel. and
.parallel..sub.x.sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0664] The criteria over the 3D SOI is below-described. The
processes 1 and 2 of method 5-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
the local region is made half. For instance, the criteria is (108)
or (108') with respect to threshold values Tratioroi or Tdiffroi.
102 ( x , y , z ) SOI ; u x i ( x , y , z ) r; 2 ( x , y , z ) SOI
; u x i - 1 ( x , y , z ) r; 2 Tratioroi ( 108 ) or ( x , y , z )
SOI ; u x i ( x , y , z ) - u x i - 1 ( x , y , z ) r; Tdiffroi , (
108 ' )
[0665] where .parallel..sub.z.sup.i(x,y,z).parallel. and
.parallel..sub.x.sup.i-1(x,y,z).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0666] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution in 3D
SOI)
[0667] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution.
The processes 1, 2 and 3 of method 5-4 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(109) or (109') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 103 ( x , y , z ) SOI ; u x i ( x , y , z ) r; 2 ( x
, y , z ) SOI ; u x i - 1 ( x , y , z ) r; 2 aboveTratioroi ( 109 )
or ( x , y , z ) SOI ; u x i ( x , y , z ) - u x i - 1 ( x , y , z
) r; aboveTdiffroi , ( 109 ' )
[0668] where .parallel..sub.x.sup.i(x,y,z).parallel. and
.sub.x.sup.i-1(x,y,z).parallel. are respectively norms (magnitudes)
of the i th and i-1 th estimates of the residual components.
[0669] Final estimate is obtained from eq. (105) or eq. (106).
[0670] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0671] [Method 5-5]
[0672] The flowchart of the method 5-5 is shown in FIG. 25. As
example, let's consider measurement of 1D displacement component
dx(x,y,z) in 3D SOI.
[0673] Possibility of divergence is detected from above-described
condition (70) or (70') in above-described process 1 of the method
5-4, and by effectively utilizing the method 5-1 based on the
method 3-1, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0674] At first, the procedure of iterative estimation is same as
that of the method 5-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0675] The process 1 of the method 5-4 is performed. (Phase
matching and estimation of the 1D residual displacement component
are performed at every point (x,y,z) in the 3D SOI.) That is, the
process 1 of the method 3-1 is performed once at every point in the
3D SOI. Moreover, using the regularization method, stably the
estimate of the 1D residual component distribution is obtained.
[0676] i-1 th estimate {haeck over (d)}.sub.x.sup.i-1(x,y,z) of 1D
displacement component distribution dx(x,y,z).
[0677] i th estimate .sub.x.sup.i-1(x,y,z) of 1D residual component
distribution u.sub.x.sup.i(x,y,z).
[0678] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 5-1 based on the method 3-1 is used.
If condition (70) or (70') is satisfied with at points or regions,
in the process 2 of the method 5-4, over sufficiently large regions
centered on the points or regions, or over the 3D SOI, the estimate
dx.sup.i(x,y,z) of the 1D displacement component dx (x,y,z) can be
3D low pass filtered or 3D median filtered as eq. (110).
{haeck over (d)}.sub.x.sup.i(x,y,z)=LPF[dx.sup.i(x,y,z)], or {haeck
over (d)}.sub.x.sup.i(x,y,z)=MED[dx.sup.i(x,y,z)] (110)
[0679] Thus, the estimation error is reduced of the residual
component.
[0680] Iterative estimation is terminated at the process 1 of the
method 5-1 based on the method 3-1, or the process 4 of the method
5-4. Hence, final estimate is obtained from eq. (67), or eq. (105),
or eq. (110).
[0681] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0682] (VI) Method 6: Measurement of 1D Displacement (One
Direction) Component Distribution in 2D ROI
[0683] [Method 6-1]
[0684] 1D x displacement component distribution in 2D ROI can be
measured by measuring 1D x displacement component distribution in
each line being parallel to x axis by means of the method 3-1, or
3-2, or 3-3, or 3-4, or 3-5 (FIG. 21).
[0685] (Process 1)
[0686] In each line being parallel to x axis in 2D ROI, the method
3-1, or 3-2, or 3-3, or 3-4, or 3-5 is utilized. The initial
estimate [eq. (57)] of the iterative estimation of the 1D
displacement component in the 2D ROI is set as zero if a priori
data is not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0687] Moreover, the methods 6-2, 6-3, 6-4, and 6-5 are
respectively baesd on the methods 3-2, 3-3, 3-4, and 3-5.
[0688] [Method 6-2]
[0689] The flowchart of the method 6-2 is shown in FIG. 22. As
example, let's consider measurement of 1D displacement component
dx(x,y) in 2D ROI. At i th estimate (i>-1), the following
processes are performed.
[0690] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution in 2D ROI)
[0691] Phase matching and estimation of the 1D residual
displacement component are performed at every point (x,y) in the 2D
ROI. That is, the processes 1 and 2 of the method 3-1 are performed
once at every point in the 2D ROI. Thus, the i th estimate of the
1D residual component distribution u.sub.x.sup.i(x,y) is obtained
as .sub.x.sup.i(x,y).
[0692] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution in 2D ROI)
[0693] The i-1 th estimate of the 1D displacement component
distribution in the 2D ROI is updated using i th estimate of the 1D
residual component distribution in the 2D ROI.
dx.sup.i(x,y)={haeck over (d)}.sub.x.sup.i-1(x,y)+.sub.x.sup.i(x,y)
(111)
[0694] Next, this estimate is 2D low pass filtered or 2D median
filter to yield the estimate of the 1D displacement component
distribution:
{haeck over (d)}.sub.x.sup.i(x,y)=LPF[dx.sup.i(x,y)], or {haeck
over (d)}.sub.x.sup.i(x,y)=MED[dx.sup.i(x,y)]. (112)
[0695] Thus, the estimation error is reduced of the residual
component compared with process 2 of the method 3-1 [eq. (63)].
Hence, phase matching of the process 1 of method 6-2 is performed
using smoothed estimate {haeck over (d)}.sub.x.sup.i(x,y) of the 1D
displacement component distribution in the 2D ROI.
[0696] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement in 2D ROI
(Condition for Making the Local Region Small)]
[0697] In order to make the spatial resolution high of the 1D
displacement component distribution measurement in the 2D ROI,
during iterative estimation, the local region used for each point
is made small, or the local region used over the 2D ROI is made
small.
[0698] The criteria for each point is below-described. The
processes 1 and 2 (method 6-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of the
local region is made half. For instance, the criteria is (113) or
(113') with respect to threshold values Tratio or Tdiff. 104 ; u x
i ( x , y ) r; ; u x i - 1 ( x , y ) r; Tratio ( 113 )
[0699] or
.parallel..sub.x.sup.i(x,y)-.sub.x.sup.i-1(x,y).parallel..ltoreq.Tdiff,
(113')
[0700] where .parallel..sub.x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0701] The criteria over the 2D ROI is below-described. The
processes 1 and 2 (method 6-2) are iteratively carried out till the
criteria is satisfied with. When the criteria is satisfied with,
the local region is made small, for instance, the length of the
local region is made half. For instance, the criteria is (114) or
(114') with respect to threshold values Tratioroi or Tdiffroi. 105
( x , y ) ROI ; u x i ( x , y ) r; 2 ( x , y ) ROI ; u x i - 1 ( x
, y ) r; 2 Tratioroi ( 114 ) or ( x , y ) ROI ; u x i ( x , y ) - u
x i - 1 ( x , y ) r; Tdiffroi , ( 114 ' )
[0702] where .parallel.hd x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0703] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution in 2D
ROI)
[0704] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution
in the 2D ROI. The processes 1, 2 and 3 of method 6-2 are
iteratively carried out till the criteria is satisfied with. For
instance, the criteria is (115) or (115') with respect to threshold
values aboveTratioroi or aboveTdiffroi. 106 ( x , y ) ROI ; u x i (
x , y ) r; 2 ( x , y ) ROI ; u x i - 1 ( x , y ) r; 2
aboveTratioroi ( 115 ) or ( x , y ) ROI ; u x i ( x , y ) - u x i -
1 ( x , y ) r; aboveTdiffroi , ( 115 ' )
[0705] where .parallel..sub.x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0706] Final estimate is obtained from eq. (111) or eq. (112).
[0707] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0708] [Method 6-3]
[0709] The flowchart of the method 6-3 is shown in FIG. 23. As
example, let's consider measurement of 1D displacement component
dx(x,y) in 2D ROI.
[0710] Possibility of divergence is detected from above-described
condition (70) or (70') in above-described process 1 of the method
6-2, and by effectively utilizing the method 6-1 based on the
method 3-1, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0711] At first, the procedure of iterative estimation is same as
that of the method 6-2 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0712] The process 1 of the method 6-2 is performed. (Phase
matching and estimation of the 1D residual displacement component
are performed at every point (x,y) in the 2D ROI.) That is, the
processes 1 and 2 of the method 3-1 are performed once at every
point in the 2D ROI. Thus, the estimate of the 1D residual
component distribution is obtained.
[0713] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 6-1 is used. If condition (70) or
(70') is satisfied with at points or regions, in the process 2 of
the method 6-2, over sufficiently large regions centered on the
points or regions, or over the 2D ROI, the estimate dx.sup.i(x,y)
[eq. (111)] of the 1D displacement component dx(x,y) can be 2D low
pass filtered or 2D median filtered as eq. (116). Thus, the
estimation error is reduced of the residual component compared with
process 2 of the method 3-1 [eq.(63)].
{haeck over (d)}.sub.x.sup.i(x,y)=LPF[dx.sup.i(x,y)], or {haeck
over (d)}.sub.i(x,y)=MED[dx.sup.i(x,y)] (116)
[0714] Thus, iterative estimation is terminated at the process 1 of
the method 6-1 based on the method 3-1, or the process 4 of the
method 6-2. Hence, final estimate is obtained from eq. (67), or eq.
(111), or eq. (116).
[0715] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0716] [Method 6-4]
[0717] The flowchart of the method 6-4 is shown in FIG. 24. As
example, let's consider measurement of 1D displacement component
dx(x,y) in 2D ROI. At i th estimate (i.gtoreq.1), the following
process 1 is performed.
[0718] (Process 1: Estimation of the 1D Residual Displacement
Component Distribution in 2D ROI)
[0719] Phase matching and estimation of the 1D residual
displacement component are performed at every point (x,y) in the 2D
ROI. That is, the process 1 of the method 3-1 is performed once at
every point in the 2D ROI.
[0720] To obtain the estimate .sub.x.sup.i(x,y) of the 1D residual
component distribution u.sup.i.sub.x(x,y) in the 2D ROI, at every
point local 1D echo cross-spectrum [eq. (59)] is evaluated from the
1D Fourier's transforms of the pre-deformation local 1D ultrasound
echo signal r.sub.1(l) and phase-matched post-deformation local 1D
ultrasound echo signal r.sup.i.sub.2 (l) Alternatively, when
pre-deformation local 1D ultrasound echo signal is phase-matched,
at every point cross-spectrum of r.sup.i.sub.1(l) and r.sub.2(l) is
evaluated.
[0721] The least squares method is implemented on the gradient of
the phase with utilization of each weight function, i.e., the
squared cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertline..sup.2, where each weight
function is normalized by the power of the cross-spectrum, i.e.,
107 l S 2 , 1 i ( l ) 2 . [ eq . ( 62 ) ]
[0722] Moreover, regularization method is also implemented. That
is, by minimizing the next functional with respect to the vector
u.sup.i comprised of the 1D residual component distribution
u.sub.x.sup.i(x,y) in the 2D ROI.
error
(u.sup.i)=.vertline..vertline.a-Fu.sup.i.vertline..vertline..sup.2+.-
alpha..sub.1i.vertline..vertline.u.sup.i.vertline..vertline..sup.2+.alpha.-
.sub.2i.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2+.alpha..sub.-
3i.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2+.alpha..su-
b.4i.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2
(117)
[0723] where
[0724] a: vector comprised of (x,y) distribution of the
cross-spectrum phase .THETA..sup.i(l) weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i(l).vertline. normalized by the magnitude
of the cross-spectrum 108 l S 2 , 1 i ( l ) 2
[0725] evaluated at every point in the 2D ROI.
[0726] F: matrix comprised of (x,y) distribution of the Fourier's
coordinate value 1 weighted with cross-spectrum
.vertline.S.sub.2,1.sup.i- (l).vertline. normalized by the
magnitude of the cross-spectrum 109 l S 2 , 1 i ( l ) 2
[0727] evaluated at every point in the 2D ROI.
[0728] .alpha..sub.1i, .alpha..sub.2i, .alpha..sub.3i,
.alpha..sub.4i: regularization parameter (at least larger than
zero)
[0729] Gu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D gradient
components of the unknown 1D residual component u.sub.x.sup.i(x,y)
110 x u x i ( x , y ) , y u x i ( x , y )
[0730] G.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D Laplacian of the
unknown 1D residual component u.sup.i(x,y) 111 2 x 2 u x i ( x , y
) + 2 y 2 u x i ( x , y )
[0731] GG.sup.TGu.sup.i: vector comprised of the finite difference
approximations of the 2D distributions of the 2D gradient
components of the 2D Laplacians of the unknown 1D residual
component u.sub.x.sup.i(x,y) 112 x ( 2 x 2 u x i ( x , y ) + 2 y 2
u x i ( x , y ) ) , y ( 2 x 2 u x i ( x , y ) + 2 y 2 u x i ( x , y
) )
[0732] As .vertline..vertline.u.sup.i.vertline..vertline..sup.2,
.vertline..vertline.Gu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.G.sup.TGu.sup.i.vertline..vertline..sup.2,
.vertline..vertline.GG.sup.TGu.sup.i.vertline..vertline..sup.2 are
positive definite, error(u.sup.i) has one minimum value. Thus, by
solving for 1D residual displacement component distribution
u.sub.x.sup.i(x,y) in the 2D ROI the simultaneous equations:
(F.sup.TF+.alpha..sub.1iI+.alpha..sub.2iG.sup.TG+.alpha..sub.3iG.sup.TGG.s-
up.TG+.alpha..sub.4iG.sup.TGG.sup.TGG.sup.TG)u.sup.i=F.sup.Ta,
(118)
[0733] estimate .sub.x.sup.i(x,y) of the 1D residual component
distribution u.sup.i.sub.x(x,y) is stably obtained. Thus,
estimation error is reduced of the residual component.
[0734] The regularization parameter of important information is set
relatively large. Thus, the regularization parameters depend on the
correlation of the local echo data (peak value of the
cross-correlation function, sharpness of the cross-correlation
function, width of the cross-correlation function), the SNR of the
cross-spectrum power, etc.; then position of the unknown
displacement component etc.
[0735] (Process 2: Update of the Estimate of the 1D Displacement
Component Distribution)
[0736] The i-1 th estimate of the 1D displacement component
distribution is updated using i th estimate of the 1D residual
component distribution.
dx.sup.i(x,y)={haeck over (d)}.sub.x.sup.i-1(x,y)+.sub.x.sup.i(x,y)
(119)
[0737] Freely, this estimate can be 2D low pass filtered or 2D
median filter to yield the estimate of the 1D displacement
component distribution.
{haeck over (d)}.sub.x.sup.i(x,y)=LPF[dx.sup.i(x,y)), or {haeck
over (d)}.sub.x.sup.i(x,y)=MED[dx.sup.i(x,y)] (120)
[0738] Hence, phase matching of the process 1 of method 6-4 is
performed using the 1D residual component data u.sub.x.sup.i(x,y)
obtained from eq. (118), or the 1D component data dx.sup.i(x,y)
obtained from eq. (119), or smoothed estimate obtained from eq.
(120).
[0739] [Process 3: Condition for Heightening the Spatial Resolution
of the 1D Displacement Component Distribution Measurement in 2D ROI
(Condition for Making the Local Region Small)]
[0740] In order to make the spatial resolution high of the 1D
displacement component distribution measurement, during iterative
estimation, the local region used for each point is made small, or
the local region used over the 2D ROI is made small.
[0741] The criteria for each point is below-described. The
processes 1 and 2 of method 6-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
the local region is made half. For instance, the criteria is (121)
or (121') with respect to threshold values Tratio or Tdiff. 113 ; u
^ x i ( x , y ) r; ; u ^ x i - 1 ( x , y ) r; T ratio ( 121 )
[0742] or
.parallel..sub.x.sup.i(x,y)-.sub.x.sup.i-1(x,y).ltoreq.Tdiff,
(121')
[0743] where .parallel..sub.x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0744] The criteria over the 2D ROI is below-described. The
processes 1 and 2 of method 6-4 are iteratively carried out till
the criteria is satisfied with. When the criteria is satisfied
with, the local region is made small, for instance, the length of
the local region is made half. For instance, the criteria is (122)
or (122') with respect to threshold values Tratioroi or Tdiffroi.
114 ( x , y ) R0I ; u ^ x i ( x , y ) r; 2 ( x , y ) R0I ; u ^ x i
- 1 ( x , y ) r; 2 T ratio roi ( 122 ) ( x , y ) R0I ; u ^ x i ( x
, y ) - u ^ x i - 1 ( x , y ) r; Tdiffroi , ( 122 ' )
[0745] where .parallel..sub.x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0746] (Process 4: Condition for Terminating the Iterative
Estimation of the 1D Displacement Component Distribution in 2D
ROI)
[0747] Below-described is the criteria for terminating the
iterative estimation of the 1D displacement component distribution.
The processes 1, 2 and 3 of method 6-4 are iteratively carried out
till the criteria is satisfied with. For instance, the criteria is
(123) or (123') with respect to threshold values aboveTratioroi or
aboveTdiffroi. 115 ( x , y ) R0I ; u ^ x i ( x , y ) r; 2 ( x , y )
R0I ; u ^ x i - 1 ( x , y ) r; 2 above T ratio roi or ( 123 ) ( x ,
y ) R0I ; u ^ x i ( x , y ) - u ^ x i - 1 ( x , y ) r; above
Tdiffroi , ( 123 ' )
[0748] where .parallel..sub.x.sup.i(x,y).parallel. and
.parallel..sub.x.sup.i-1(x,y).parallel. are respectively norms
(magnitudes) of the i th and i-1 th estimates of the residual
components.
[0749] Fidnal estimate is obtained from eq. (119) or eq. (120).
[0750] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0751] [Method 6-5]
[0752] The flowchart of the method 6-5 is shown in FIG. 25. As
example, let's consider measurement of 1D displacement component
dx(x,y) in 2D ROI.
[0753] Possibility of divergence is detected from above-described
condition (70) or (70') in above-described process 1 of the method
6-4, and by effectively utilizing the method 6-1 based on the
method 3-1, even if the SNR of the ultrasound echo signals are low,
accurate 1D displacement component measurement can be realized.
[0754] At first, the procedure of iterative estimation is same as
that of the method 6-4 (Processes 1, 2, 3, and 4). At i th
estimate, the following processes are performed.
[0755] The process 1 of the method 6-4 is performed. (Phase
matching and estimation of the 1D residual displacement component
are performed at every point (x,y) in the 2D ROI.) That is, the
process 1 of the method 3-1 is performed once at every point in the
2D ROI. Moreover, using the regularization method, stably the
estimate of the 1D residual component distribution is obtained.
[0756] i-1 th estimate {haeck over (d)}.sub.x.sup.i-1(x,y) of 1D
displacement component distribution dx(x,y).
[0757] i th estimate .sub.x.sup.i(x,y) of 1D residual component
distribution u.sub.x.sup.i(x,y).
[0758] During this estimation, if neither condition (70) nor (70')
is satisfied with, the method 6-1 based on the method 3-1 is used.
If condition (70) or (70') is satisfied with at points or regions,
in the process 2 of the method 6-4, over sufficiently large regions
centered on the points or regions, or over the 2D ROI, the estimate
dx.sup.i(x,y) of the 1D displacement componentdx(x,y) can be 2D low
pass filtered or 2D median filtered as eq. (124).
{haeck over (d)}.sub.x.sup.i(x,y)=LPF[dx.sup.i(x,y)], or {haeck
over (d)}.sub.x.sup.i(x,y)=MED[dx.sup.i(x,y)] (124)
[0759] Thus, the estimation error is reduced of the residual
component.
[0760] Iterative estimation is terminated at the process 1 of the
method 6-1 based on the method 3-1, or the process 4 of the method
6-4. Hence, final estimate is obtained from eq. (67), or eq. (119),
or eq. (124).
[0761] The initial estimate [eq. (57)] of the iterative estimation
of the 1D displacement component is set as zero if a priori data is
not given about displacements of body motion nor applied
compression. Alternatively, values accurately estimated at
neighborhood can be used (high correlation or least squares).
[0762] In the 3D SOI, the 3D displacement vector distribution can
also be measured using method 4 or method 5 by changing the adapted
direction. In the 2D ROI, the 2D displacement vector distribution
can also be measured using method 6 by changing the adapted
direction. Except for threshold value for terminating the iterative
estimation, other threshold values can be updated. Estimate can
also be performed non-iteratively.
[0763] When applying the regularization method, in addition to the
magnitude of the unknown displacement vector, spatial continuity
and differentiability of the unknown displacement vector
distribution, mechanical properties of tissue (e.g.,
incompressibility), and compatibility conditions of displacement
vector distribution and displacement component distribution, as the
a priori knowledge, used is temporal continuity and
differentiability of the unknown series of the displacement vector
distribution and displacement component distribution. The
regularization parameter depends on time-space dimension number,
direction of the unknown displacement component, position of the
unknown displacement vector, time, etc.
[0764] Thus, as the displacement vector can be measured accurately,
consequently, in addition to 3D strain tensor, accurately measured
can be 2D strain tensor, one strain component, 3D strain rate
tensor, 2D strain rate tensor, one strain rate component,
acceleration vector, velocity vector, etc.
[0765] (VII) Differential Filter
[0766] The strain tensor components can be obtained by spatial
differential filtering with suitable cut off frequency in time
domain or frequency domain the measured 3D, or 2D displacement
vector components, or measured 1D direction displacement component
in the 3D, 2D, or 1D ROI. The strain rate tensor components,
acceleration vector components, or velocity vector components can
be obtained by time differential filtering with suitable cut off
frequency in time domain or frequency domain the measured time
series of displacement components, or strain components. The strain
rate tensor components can be obtained from the strain tensor
components directly measured by below-described signal
processing.
[0767] As above described, when measuring displacement from the
gradient of the echo cross-spectrum phase, to result the more
accurate measurement accuracy, the least squares method can be
applied with utilization as the weight function of the squares of
the cross-spectrum usually normalized by the cross-spectrum power,
where, to stabilize the measurement, the regularization method can
be applied, by which a priori information can be incorporated,
i.e., about within the ROI the magnitude of the unknown
displacement vector, spatial continuity and differentiability of
the unknown displacement vector distribution etc.
[0768] Next, as the estimation methods of the displacements during
the iterative estimation to update 3D, 2D, 1D displacement
component, in order to reduce calculation amount and shorten
calculation time, other methods are also described. These
estimation methods can be used in combination, or one of them can
be used. To realize real-time measurement, the estimate can also be
performed non-iteratively.
[0769] In order to reduce calculation amount and shorten
calculation time, calculation process is simplified. That is, as
the cross-spectrum phase .theta. (.omega.x, .omega.y, .omega.z) is
represented as .theta..sub.2(.omega.x, .omega.y,
.omega.z)-.theta..sub.1(.omega.x, .omega.y, .omega.z) using the
phases .theta..sub.1(.omega.x, .omega.y, .omega.z) and
.theta..sub.2 (.omega.x, .omega.y, .omega.z) respectively obtained
from 3D Fourier's transforms R.sub.1(.omega.x, .omega.y, .omega.z)
and R.sub.2(.omega.x, .omega.y, .omega.z) of the local echo signals
under pre- and post-deformation, the displacement vector
u(=(ux,uy,uz).sup.T) is represented as 116 ( ux uy uz ) = grad (
arg [ R 2 * ( x , y , z ) R 1 ( x , y , z ) ] ) ( where grad = ( x
, y , z ) r ) = ( x ( x , y , z ) y ( x , y , z ) z ( x , y , z ) )
= ( x ( 2 ( x , y , z ) - 1 ( x , y , z ) ) y ( 2 ( x , y , z ) - 1
( x , y , z ) ) z ( 2 ( x , y , z ) - 1 ( x , y , z ) ) ) = Im [
grad ( ln { R 2 * ( x , y , z ) R 1 ( x , y , z ) } ) ] = ( - Re [
R 2 ( x , y , z ) ] x Im [ R 2 ( x , y , z ) ] - x Re [ R 2 ( x , y
, z ) ] Im [ R 2 ( x , y , z ) ] R 2 ( x , y , z ) 2 + Re [ R 1 ( x
, y , z ) ] x Im [ R 1 ( x , y , z ) ] - x Re [ R 1 ( x , y , z ) ]
Im [ R 1 ( x , y , z ) ] R 1 ( x , y , z ) 2 - Re [ R 2 ( x , y , z
) ] y Im [ R 2 ( x , y , z ) ] - y Re [ R 2 ( x , y , z ) ] Im [ R
2 ( x , y , z ) ] R 2 ( x , y , z ) 2 + Re [ R 1 ( x , y , z ) ] y
Im [ R 1 ( x , y , z ) ] - y Re [ R 1 ( x , y , z ) ] Im [ R 1 ( x
, y , z ) ] R 1 ( x , y , z ) 2 - Re [ R 2 ( x , y , z ) ] z Im [ R
2 ( x , y , z ) ] - z Re [ R 2 ( x , y , z ) ] Im [ R 2 ( x , y , z
) ] R 2 ( x , y , z ) 2 + Re [ R 1 ( x , y , z ) ] z Im [ R 1 ( x ,
y , z ) ] - z Re [ R 1 ( x , y , z ) ] Im [ R 1 ( x , y , z ) ] R 1
( x , y , z ) 2 ) .
[0770] Then, using the phases of the high SNR frequency the
displacement vector u can be obtained by partial differentiating in
the frequency directions .omega.x, .omega.y, .omega.z the
difference between the phases .theta..sub.2 (.omega.x, .omega.y,
.omega.z) and .theta..sub.1 (.omega.x, .omega.y, .omega.z), or by
calculating the difference between partial differentiations in the
frequency directions .omega.x, .omega.y, .omega.z of the phases
.theta..sub.2 (.omega.x, .omega.y, .omega.z) and
.theta..sub.1(.omega.x, .omega.y, .omega.z), or by using Fourier's
transform values Re[R.sub.2(.omega.x, .omega.y, .omega.z)],
Im[R.sub.2(.omega.x, .omega.y, .omega.z)], Re[R.sub.1(.omega.x,
.omega.y, .omega.z)], Im[R.sub.1(.omega.x, .omega.y, .omega.z)],
and their partial derivatives in the frequency directions .omega.x,
.omega.y, .omega.z without unwrapping the phase. These partial
derivatives can be obtained by finite-difference approximating or
differential filtering. Freely, the phases, the signal components,
or numerator and denominator can be moving averaged or low-pass
filtered in the frequency domain. The final estimate can be the
mean vector calculated from the displacement data obtained at high
SNR frequencies.
[0771] The 2D displacement vector and one direction displacement
component can respectively be obtained in a similar way by
calculating 2D and 1D Fourier's transforms.
[0772] The simultaneous equations of the above-described equation
can be solved in the frequency domain, or spatial and temporal
simultaneous equations of the above-described equation can be
handled, where above-described regularization method can be
applied.
[0773] When performing 1D (one direction) calculation, in order to
reduce calculation amount and shorten calculation time, calculation
process is simplified. That is, for instance, when performing x
direction calculation, as the cross-spectrum phase .theta.
(.omega.x, .omega.y, .omega.z) is represented as
.theta.(.omega.x)=ux.multidot..omega.x, the displacement is
obtained form the phase of the high SNR frequency (y direction
calculation; .theta.(.omega.y)=uy.multidot..omega.y, z direction
calculation; .theta.(.omega.z)=uz.multidot..omega.z). The final
estimate can be the mean value calculated from the displacement
data obtained at high SNR frequencies.
[0774] When large displacement needs to be handled, before
estimating the gradient of the cross-spectrum phase, the phase was
unwrapped, or the displacement was coarsely estimated by
cross-correlation method. Thus, measurement procedure had become
complex one. Otherwise, to cope with these complexity, the
measurement procedure is made simpler without these processes by
introducing process of thinning out data and remaking data interval
original.
[0775] The simultaneous equations of the above-described equation
can be solved in the frequency domain, or spatial and temporal
simultaneous equations of the above-described equation can be
handled, where above-described regularization method can be
applied.
[0776] Otherwise, echo signals are acquired at two different time,
freely, the aouto-correlation method (beam direction or scan
direction) and the regularization method can be equipped.
[0777] Otherwise, freely, ultrasound Doppler's method can be
equipped. The Doppler's shift can be detected in beam direction or
scan direction.
[0778] There are many methods for detecting the Doppler's shift.
From the phase distribution .theta..sub.ZR(x,y,z,t)=tan.sup.-1
(Re[Z.sub.R(x,y,z,t)]/Im[Z.sub.R(x,y,z,t)]) of the quadrate
demodulation signal Z.sub.R(x,y,z,t)
(=Re[Z.sub.R(x,y,z,t)]+jIm[Z.sub.R(x,y,z,t)] in R axis direction
acquired at each position (x,y,z) in the ROI, for instnce, the
velocity component vx in x axis direction (R=x) at time t=T and at
position (X,Y,Z) can be obtained as 117 vx = - 1 s x t [ c x f 0 x
tan - 1 ( Re [ Z x ( x , y , z , t ) ] Im [ Zx ( x , y , z , t ) ]
) ] | x = X , y = Y , z = Z , t = T = - 1 s x ( c x f 0 x Re [ Zx (
x , y , z , t ) ] t [ Im [ Zx ( x , , y , z , t ) ] ] - t [ Re [ Zx
( x , y , z , t ) ] ] Im [ Zx ( x , , y , z , t ) ] Re [ Zx ( x , y
, z , t ) ] 2 + Im [ Zx ( x , y , z , t ) ] 2 ) | x = X , y = Y , z
= Z , t = T .
[0779] c.sub.R is the ultrasound propagating velocity and scan
velocity respectively when R axis is beam axis and scan axis.
f.sub.0R is ultrasound carrier frequency and sine frequency
respectively when R axis is beam axis and scan axis. s.sub.R is 4.0
and 2.0 respectively when R axis is beam axis and scan axis. As
above-described, temporal gradient of the phase .theta..sub.ZR
(x,y,z,t) can also be obtained by finite difference approximating
or differential filtering after obtaining the phase .theta..sub.ZR
(x,y,z,t). Freely, the phases, the signal components, or numerator
and denominator can be moving averaged or low-pass filtered in the
time domain. Thus, the velocity component distributions (series)
can be obtained in the ROI.
[0780] The spatial and temporal simultaneous equations of the
above-described equation can be handled, where above-described
regularization method can be applied.
[0781] By multiplying pulse transmitting interval Ts to each
velocity component distributions (series), the displacement
component distribution (series) can be obtained. Alternatively, by
integrating the velocity vector component distributions (series),
the displacement vector distribution (series) can be obtained.
[0782] From temporal spatial derivatives of these velocity vector
component distributions (series) or displacement vector component
distributions (series),obtained are strain tensor component
distributions (series),acceleration vector component distributions
(series), and strain rate tensor component distributions
(series).
[0783] Otherwise, freely, a method for directly obtaining strain
tensor components can be equipped, i.e., from spatial partial
derivative of the phase of the quadrate demodulate signal (beam
direction or scan direction) of the ultrasound echo signals.
[0784] From the phase distribution .theta..sub.ZR
(x,y,z,t)=tan.sup.-1 (Re[Z.sub.R(x,y,z,t)]/Im[Z.sub.R(x,y,z,t)]) of
the quadrate demodulation signal Z.sub.R(x,y,z,t)
(=Re[Z.sub.R(x,y,z,t)]+jIm Z.sub.R(X, y, Z, t)] in R axis direction
acquired at each position (x,y,z) in the ROI, for instance, the
normal strain component .epsilon. xx in x axis direction (R=x) at
time t=T and at position (X,Y,Z) can be obtained as 118 xx ( X , Y
, Z , T ) = x u x ( x , y , z , t ) x = X , y = Y , z = Z , t = T =
- 1 s x x [ c x f 0 x tan - 1 ( Re [ Zx ( x , y , z , t ) ] Im [ Zx
( x , y , z , t ) ] ) ] | x = X , y = Y , z = Z , t = T = - 1 s x (
c x f 0 x Re [ Zx ( x , y , z , t ) ] x [ Im [ Zx ( x , , y , z , t
) ] ] - x [ Re [ Zx ( x , y , z , t ) ] ] Im [ Z x ( x , , y , z ,
t ) ] Re [ Zx ( x , y , z , t ) ] 2 + Im [ Zx ( x , y , z , t ) ] 2
) | x = X , y = Y , z = Z , t = T .
[0785] c.sub.R is the ultrasound propagating velocity and scan
velocity respectively when R axis is beam axis and scan axis.
f.sub.0R is ultrasound carrier frequency and sine frequency
respectively when R axis is beam axis and scan axis. s.sub.R is 4.0
and 2.0 respectively when R axis is beam axis and scan axis. As
above-described, spatial gradient of the phase .theta..sub.ZR
(x,y,z,t) can also be obtained by finite difference approximating
or differential filtering after obtaining the phase .theta..sub.ZR
(x,y,z,t). Freely, the phases, the signal components, or numerator
and denominator can be moving averaged or low-pass filtered in the
space domain. For instance, the shear strain component .epsilon.xy
in x-y plane (R=x and y) at time t=T and at position (X,Y,Z) can be
obtained as 119 xx ( X , Y , Z , T ) = 1 2 ( x uy ( x , y , z , t )
+ y ux ( x , y , z , t ) ) | x = X , y = Y , z = Z , t = T = 1 2 (
- 1 s y x [ c y f 0 y tan - 1 ( Re [ Zy ( x , y , z , t ) ] Im [ Zy
( x , y , z , t ) ] ) ] - 1 s x y [ c x f 0 x tan - 1 ( Re [ Zx ( x
, y , z , t ) ] Im [ Zx ( x , y , z , t ) ] ) ] ) | x = X , y = Y ,
z = Z , t = T = 1 2 ( - 1 s y ( c y f 0 y Re [ Zy ( x , y , z , t )
] x [ Im [ Zy ( x , , y , z , t ) ] ] - x [ Re [ Zy ( x , y , z , t
) ] ] Im [ Zy ( x , y , z , t ) ] Re [ Zy ( x , y , z , t ) ] 2 +
Im [ Zy ( x , y , z , t ) ] 2 ) - 1 s x ( c x f 0 x Re [ Zx ( x , y
, z , t ) ] y [ Im [ Zx ( x , , y , z , t ) ] ] - y [ Re [ Zx ( x ,
y , z , t ) ] ] Im [ Z x ( x , y , z , t ) ] Re [ Zx ( x , y , z ,
t ) ] 2 + Im [ Zx ( x , y , z , t ) ] 2 ) ) | x = X , y = Y , z = Z
, t = T .
[0786] Thus, the strain component distributions (series) can be
obtained in the ROI.
[0787] The spatial and temporal simultaneous equations of the
above-described equation can be handled, where above-described
regularization method can be applied.
[0788] By integrating the partial derivatives of displacement
vector component distributions (series), the displacement vector
distribution (series) can be obtained.
[0789] From temporal spatial derivatives of these strain tensor
component distributions (series) or displacement vector component
distributions (series), obtained are strain rate tensor component
distributions (series), and acceleration vector component
distributions (series).
[0790] Freely, (I-1) complex cross-correlation method (phase in
beam direction or scan direction of complex cross-correlation
function signal obtained from complex analytic signals or quadrate
detection signals, or obtained from cross-correlation of ultrasound
echo signals) is utilized, or (I-2) both of complex
cross-correlation method (beam direction or scan direction) and the
regularization method are utilized, or (I-3) at least 2D
distribution (including beam direction or not) of the phase of 3D,
or 2D, or 1D complex cross-correlation function signal and the
regularization method are utilized. That is, methods (I-1), (I-2),
and (I-3) can be equipped.
[0791] On the method (I-3), for instance, the next equation holds
for unknown 3D displacement vector (ux,uy,uz).sup.T at each point
(X,Y,Z) at time t=T: 120 cc ( 0 , 0 , 0 ) + x cc ( x , y , z ) | x
= 0 , y = 0 , z = 0 ux + y cc ( x , y , z ) | x = 0 , y = 0 , z = 0
uy + z cc ( x , y , z ) | x = 0 , y = 0 , z = 0 uz = 0.
[0792] .theta..sub.cc(X,Y,Z;x,y,z) is the 3D phase distribution
(x,y,z) of the complex cross-correlation function Cc(X,Y,Z;x,y,z)
of the point (X,Y,Z) evaluated from rf echo signals with respect to
transmitted ultrasound pulses at the time t=T and t=T+.DELTA.T:
.theta..sub.cc(X,Y,Z;x,y,z)=tan.sup.-1(Re[Cc(X,Y,Z;x,y,z)]/Im[Cc(X,Y,Z;x,y-
,z)]),
[0793] where the coordinate (x,y,z) has the origin at (X,Y,Z). In
the SOI, occasionally also in time direction, these equations hold
(simultaneous equations), and can be solved by least squared
method, where, freely, regularization method can be applied (the
temporal and spatial magnitude of the unknown displacement vector
distribution, temporal and spatial continuity and differentiability
of the unknown displacement vector distribution). Thus, the
displacement vector distribution (series) can be obtained. The
gradients of the phase .theta..sub.cc (X,Y,Z;x,y z) can be obtained
by finite difference approximating or differential filtering.
However, for instance, x partial derivative
.differential./.differential.-
x.multidot..theta..sub.cc(x,y,z).vertline..sub.x=0,y=0,z=0 can be
obtained as:
{Re[Cc(X,Y,Z;0,0,0)].times..differential./.theta.x.multidot.Im[Cc(X,Y,Z;x,-
y,z)].vertline..sub.x=0, y=0,
z=0-.differential./.theta.x.multidot.Re[Cc(X-
,Y,Z;x,y,z)].vertline..sub.x=0, y=0,
y=0.times.Im[Cc(X,Y,Z;0,0,0)]}/{Re[Cc-
(X,Y,Z;0,0,0)].sup.2+Im[Cc(X,Y,Z;0,0,0)].sup.2}.
[0794]
.theta./.theta.x.multidot.Re[Cc(X,Y,Z;x,y,z)].vertline..sub.x=0,
y=0, z=0 can be obtained by finite difference approximating or
differential filtering. Freely, the phases, the signal components,
or numerator and denominator can be moving averaged or low-pass
filtered in the time domain.
[0795] For instance, the next equation holds for unknown 2D
displacement vector (ux,uy).sup.T at each point (X,Y,Z) at time
t=T: 121 cc ( 0 , 0 ) + x cc ( x , y ) | x = 0 , y = 0 ux + y cc (
x , y ) | x = 0 , y = 0 uy = 0.
[0796] .theta..sub.cc (X,Y,Z;x,y) is the 2D phase distribution
(x,y) of the complex cross-correlation function Cc(X,Y,Z;x,y) of
the point (X,Y,Z) evaluated from rf echo signals with respect to
transmitted ultrasound pulses at the time t=T and t=T+.DELTA.T.
[0797] On the method (I-1), utilized is phase of complex
cross-correlation function signal in beam direction or scan
direction. The next equation holds for unknown displacement
component ux at each point (X,Y,Z) at time t=T (the
auto-correlation function method's equation): 122 cc ( 0 ) + x cc (
x ) | x = 0 ux = 0.
[0798] .theta..sub.cc (X,Y,Z;x) is the 1D phase distribution (x) of
the complex cross-correlation function Cc(X,Y,Z;x) of the point
(X,Y,Z) evaluated from rf echo signals with respect to transmitted
ultrasound pulses at the time t=T and t=T+.DELTA.T.
[0799] In the ROI, by solving this equation for unknown
displacement component ux at each point, the displacement component
distribution (series) can be obtained.
[0800] On the method (I-2), in the ROI, occasionally also in time
direction, this equation holds in beam direction or scan direction,
and the derived simultaneous equations can be solved by least
squared method, where, freely, regularization method can be applied
(the temporal and spatial magnitude of the unknown displacement
component distribution, temporal and spatial continuity and
differentiability of the unknown displacement component
distribution). Thus, the displacement component distribution
(series) can be obtained.
[0801] On the methods (I-3) and (I-2), the unknown displacement
vector and the unknown displacement component can be dealt as
locally uniform ones. That is, under the assumption that the local
region uniformly moves, the simultaneous equations hold for the
unknown local displacement vector or the unknown local displacement
component. Otherwise, the simultaneous equations hold under the
assumption that the displacement is uniform for temporal finite
interval. Thus, the spatial distribution (series) can be
obtained.
[0802] The next method can be equipped. That is, the strain tensor
component can be directly obtained from spatial derivative of the
no time delay phase
.theta..sub.cc(x,y,z;0,0,0)=tan.sup.-1(Re[Cc(x,y,z;0,0,0)]/Im-
[Cc(x,y,z;0,0,0)]) of the 3D complex cross-correlation function, of
the no time delay phase
.theta..sub.cc(x,y,z;0,0)=tan.sup.-1(Re[Cc(x,y,z;0,0)]/I-
m[Cc(x,y,z;0,0)]) of the 2D complex cross-correlation function
(including beam direction or not), or of the no time delay phase
.theta..sub.cc(x,y,
z;0)=tan.sup.-1(Re[Cc(x,y,z;0)]/Im[Cc(x,y,z;0)]) of the 1D complex
cross-correlation function (beam direction or scan direction) of
the point (x,y,z) evaluated from rf echo signals with respect to
transmitted ultrasound pulses at the time t=T and t=T+.DELTA.T.
[0803] For instance, the normal strain component E xx in x axis
direction (R=x) at time t=T and at position (X,Y,Z) can be obtained
as: 123 xx ( X , Y , Z , T ) = x u x ( x , y , z , t ) x = X , y =
Y , z = Z , t = T = - 1 s R x [ c x f 0 x cc ( x , y , z , t ) ] |
x = X , y = Y , z = Z , t = T = - 1 s R ( c x f 0 x Re [ Cc ( x , y
, z , t ) ] x [ Im [ Cc ( x , , y , z , t ) ] ] - x [ Re [ Cc ( x ,
y , z , t ) ] ] Im [ Cc ( x , , y , z , t ) ] Re [ Cc ( x , y , z ,
t ) ] 2 + Im [ cc ( x , y , z , t ) ] 2 ) | x = X , y = Y , z = Z ,
t = T .
[0804] c.sub.R is the ultrasound propagating velocity and scan
velocity respectively when R axis is beam axis and scan axis.
f.sub.0R is ultrasound carrier frequency and sine frequency
respectively when R axis is beam axis and scan axis. s.sub.R is 4.0
and 2.0 respectively when R axis is beam axis and scan axis. As
above-described, spatial gradient of the phase
.theta..sub.cc(x,y,z,t) can also be obtained by finite difference
approximating or differential filtering after obtaining the phase
.theta..sub.cc(x,y,z,t). Freely, the phases, the signal components,
or numerator and denominator can be moving averaged or low-pass
filtered in the space domain. Thus, the strain component
distributions (series) can be obtained in the ROI.
[0805] The spatial and temporal simultaneous equations of the
above-described equation can be handled, where above-described
regularization method can be applied.
[0806] By integrating the partial derivatives of displacement
vector component distributions (series), the displacement vector
distribution (series) can be obtained.
[0807] From temporal spatial derivatives of these strain tensor
component distributions (series) or displacement vector component
distributions (series), obtained are strain rate tensor component
distributions (series), and acceleration vector component
distributions (series).
[0808] Freely, (II-1) complex quadrate detection signal method
(beam direction or scan direction) is utilized, or (II-2) both of
complex quadrate detection signal method (beam direction or scan
direction) and the regularization method are utilized, or (II-3) at
least 2D distribution (including beam direction or not) of the
phase of complex quadrate detection signal and the regularization
method are utilized (Optical flow algorithm is applied to the phase
of the quadrate detection signal.). That is, methods (II-1),
(II-2), and (II-3) can be equipped.
[0809] On the method (II-3), for instance, the next equation holds
for unknown 3D displacement vector (ux,uy,uz).sup.T at each point
(X,Y,Z) at time t=T: 124 x A ( x , y , z , t ) x = X , y = Y , z =
Z , t = T ux + y A ( x , y , z , t ) x = X , y = Y , z = Z , t = T
uy + z A ( x , y , z , t ) x = X , y = Y , z = Z , t = T uz + t A (
x , y , z , t ) x = X , y = Y , z = Z , t = T t = 0
[0810] (or for unknown 3D velocity vector (vx,vy,vz).sup.T: 125 x A
( x , y , z , t ) x = X , y = Y , z = Z , t = T vx + y A ( x , y ,
z , t ) x = X , y = Y , z = Z , t = T vy + z A ( x , y , z , t ) x
= X , y = Y , z = Z , t = T vz + t A ( x , y , z , t ) x = X , y =
Y , z = Z , t = T = 0
[0811] ).
[0812] .theta..sub.A(x,y,z,t) is the 3D phase distribution (x,y,z)
of the complex quadrate detection signal A(x,y,z,t)
(=Re[A(x,y,z,t)]+jIm[A(x,y,z- ,t)] of the point (x,y,z) at the time
t (.DELTA.t: transmitted pulse interval):
[0813] .theta..sub.A
(x,y,z,t)=tan.sup.-1(Re[.DELTA.(x,y,z,t)]/Im[A(x,y,z,- t)]) In the
SOI, occasionally also in time direction, these equations hold
(simultaneous equations), and can be solved by least squared
method, where, freely, regularization method can be applied [the
temporal and spatial magnitude of the unknown displacement
(velocity) vector distribution, temporal and spatial continuity and
differentiability of the unknown displacement (velocity) vector
distribution]. Thus, the displacement (velocity) vector
distribution (series) can be obtained. The temporal and spatial
gradients of the phase .theta..sub.A (x,y,z,t) can be obtained by
finite difference approximating or differential filtering. However,
for instance, x partial derivative .differential./.theta.x.multi-
dot..theta..sub.A(x,y,z,t).vertline..sub.x=X,y=Y,z=Z,t=T can be
obtained as
{Re[A(X,Y,Z,T)].times..differential./.differential.x.multidot.Im[A(x,y-
,z,t)].vertline..sub.x=X, y=Y, z=Z,
t=T-.theta./.theta.x.multidot.Re[A(x,y- ,z,t)].vertline..sub.x=X,
y=Y, z=Z, t=T.times.Im[A(X,Y,Z,T)]}/{Re[A(X,Y,Z;-
T)].sup.2+Im[A(X,Y,Z,T)].sup.2}.
[0814] .theta./.theta.x.multidot.Re[A(x,y,z,t)].vertline..sub.x=X,
y=Y, z=Z, t=T can be obtained by finite difference approximating or
differential filtering. Freely, the phases, the signal components,
or numerator and denominator can be moving averaged or low-pass
filtered in the time domain.
[0815] For instance, the next equation holds for unknown 2D
displacement vector (ux,uy).sup.T at each point (X,Y,Z) at time
t=T: 126 x A ( x , y , t ) x = X , y = Y , t = T ux + y A ( x , y ,
t ) x = X , y = Y , t = T uy + t A ( x , y , t ) x = X , y = Y , t
= T t = 0
[0816] (or for unknown 2D velocity vector (vx,vy).sup.T: 127 x A (
x , y , t ) x = X , y = Y , t = T vx + y A ( x , y , t ) x = X , y
= Y , t = T vy + t A ( x , y , t ) x = X , y = Y , t = T = 0
[0817] ).
[0818] .theta..sub.A (x,y,t) is the 2D phase distribution (x,y) of
the complex quadrate detection signal A(x,y,t)
(=Re[A(x,y,t)]+jIm[A(x,y,t)] of the point (x,y) at the time t
(.DELTA.t: transmitted pulse interval):
.theta..sub.A
(x,y,z,t)=tan.sup.-1(Re[A(x,y,z,t)]/Im[A(x,y,z,t)]).
[0819] On the method (II-1), utilized is phase of complex quadrate
detection signal in beam direction or scan direction. The next
equation holds for unknown displacement component ux at each point
(X,Y,Z) at time t=T: 128 x A ( x , t ) x = X , t = T ux + t A ( x ,
t ) x = X , t = T t = 0
[0820] (or for unknown velocity component vx (the Doppler's
equation): 129 x A ( x , t ) x = X , t = T vx + t A ( x , t ) x = X
, t = T = 0
[0821] ).
[0822] .theta..sub.A (x, t) is the 1D phase distribution (x) of the
complex quadrate detection signal A(x,t) of the point (x) at the
time t (.DELTA.T: transmitted pulse interval).
[0823] In the ROI, by solving this equation for unknown
displacement component ux (unknown velocity component vx) at each
point, the displacement (velocity) component distribution (series)
can be obtained.
[0824] On the method (II-2), in the ROI, occasionally also in time
direction, this equation holds in beam direction or scan direction,
and the derived simultaneous equations can be solved by least
squared method, where, freely, regularization method can be applied
(the temporal and spatial magnitude of the unknown displacement
component distribution, temporal and spatial continuity and
differentiability of the unknown displacement component
distribution). Thus, the displacement component distribution
(series) can be obtained.
[0825] On the methods (II-3) and (II-2), the unknown displacement
(velocity) vector and the unknown displacement (velocity) vector
component can be dealt as locally uniform ones. That is, under the
assumption that the local region uniformly moves, the simultaneous
equations hold for the unknown local displacement (velocity) vector
or the unknown local displacement (velocity) component. Otherwise,
the simultaneous equations hold under the assumption that the
displacement (velocity) is uniform for temporal finite interval.
Thus, the spatial distribution (series) can be obtained.
[0826] The displacement vector distribution (series) can also be
obtained by integrating the obtained velocity vector component
distributions (series), or by multiplying transmitted pulse
interval Ts to the obtained velocity vector component distributions
(series).
[0827] From temporal spatial derivatives of these velocity vector
distribution (series) or displacement vector distribution (series),
obtained are strain tensor component distributions (series), strain
tensor rate component distributions (series), and acceleration
vector component distributions (series).
[0828] There are other various methods for estimating remaining
estimation error vector. These methods can also be utilized in the
same way. When estimation error of the displacement vector or
remaining estimation error vector is detected during the iterative
estimation a priori as the point of time-space magnitude and
time-space continuity, for instance, the estimate can be cut by
compulsion such that the estimate ranges from the given smallest
value to the given largest value, or such that the difference
between the estimates of the neighboring points settle within the
given ranges.
[0829] As explained, on this conduct form, by iterative estimation
the measurement accuracy can be improved of the displacement vector
in the 3D SOI, particularly, 3D displacement vector, obtained from
the cross-spectrum phase gradient etc. of the ultrasound echo
signals acquired as the responses to more than one time transmitted
ultrasound. The local echo signal can be shifted by multiplying
complex exponential, or interporation can be performed after
shifting sampling ultrasound signal. The present invention can
improve measurement accuracy of lateral displacements (orthogonal
directions to beam direction). Furthermore, the present invention
can simplify calculation process into one without unwrapping the
cross-spectrum phase nor utilizing cross-correlation method in
order to reduce calculation amount and shorten calculation
time.
[0830] Moreover, on this conduct form, large displacement (vector)
and large strain (tensor) can be accurately measured by tracking
ultrasound echo signals of targeted tissue using echo signal phase
as the index (the local echo signal can be shifted by multiplying
complex exponential, or interporation can be performed after
shifting sampling ultrasound signal.), and by adding successively
estimated at least more than two displacements.
[0831] Furthermore, on this conduct form, elastic constant and
visco elastic constant can be accurately measured with high freedom
of configurations of displacement (strain) sensors, mechanical
sources, reference regions (mediums).
[0832] Next, explains are about elasticity and visco-elasticity
constants measurement apparatus related to one of conduct forms of
the present invention. The elasticity and visco-elasticity
constants measurement apparatus related to this conduct form
utilize the apparatus shown in FIG. 1 (same as that of the
above-explained displacement vector and strain measurement), and
the apparatus measures elastic constants and visco elastic
constants from displacement vector, strain tensor, etc. measured by
using the above-explained displacement and strain measurement
method.
[0833] At first, the assumptions are explained of the elasticity
and visco-elasticity constants measurement apparatus related to
this conduct form. The following constants are assumed to be
measured only in the target ROI (SOI) set in the measurement
object, elastic constants such as shear modulus, Poisson's ratio,
etc., visoc elastic constants such as visco shear modulus, visco
Poisson's ratio, etc., delay times or relaxation times relating
these elastic constants and visco elastic constants, or density.
All the mechanical sources are assumed to exist outside of the ROI.
Then, if there exist other mechanical sources in addition to set
mechanical sources or if the mechanical sources are uncontrollable,
the following constants can be measured in the target ROI (SOI),
elastic constants such as shear modulus, Poisson's ratio, etc.,
visoc elastic constants such as visco shear modulus, visco
Poisson's ratio, etc., delay times or relaxation times relating
these elastic constants and visco elastic constants, or density.
Neither information is needed about mechanical sources, such as
positions, force directions, force magnitudes, etc. Moreover
neither stress data nor strain data are needed at the target body
surface. Only the ROI is modeled using finite difference method or
finite element method.
[0834] If the mechanical sources originally exist near the ROI,
only the mechanical sources can be utilized. In the case of
observation of living tissues, for instance, such mechanical
sources include normally uncontrollable mechanical sources such as
heart motion, respiratory motion, blood vessel motion, body motion.
In general, lung, air, blood vessel, blood are included in the ROI.
In this case, without disturbing the deformation field, the
following constants can be measured, i.e., elastic constants such
as shear modulus, Poisson's ratio, etc., visoc elastic constants
such as visco shear modulus, visco Poisson's ratio, etc., delay
times or relaxation times relating these elastic constants and
visco elastic constants, or density. This is effective particularly
when the ROI deeply situates.
[0835] When solving the first order partial differential equations,
as initial conditions the following can be utilized, i.e.,
reference shear modulus and reference Poisson's ratio for elastic
constants, reference visco shear modulus and reference visco
Poisson's ratio for visco elastic constants, reference density for
density. In this case, reference mediums or reference regions are
set in the original ROI or near the original ROI, after which the
final ROI is set such that the final ROI includes the original ROI
as well as the references. By measuring in the ROI including
reference regions strain tensor field, strain rate tensor field,
and acceleration vector field, the references are realized.
[0836] The size and the position of the reference mediums or
reference regions should be set such that they should widely cross
the direction of the dominant tissue deformation. For instance, if
the mechanical source has large contact surface, large reference
region must be set. Otherwise, if the mechanical source has small
contact surface, by setting the reference region near the
mechanical source, small reference region can be used. The
estimates can be also used as their references.
[0837] The present invention can provide absolute shear modulus
distribution, relative shear modulus distribution, absolute
Poisson's ratio distribution, relative Poisson's ratio
distribution, absolute visco shear modulus distribution, relative
visco shear modulus distribution, absolute visco Poisson's ratio
distribution, relative visco Poisson's ratio distribution, absolute
or relative delay time distributions relating these elastic
constants and visco elastic constants, or absolute or relative
relaxation time distributions relating these elastic constants and
visco elastic constants, absolute density distribution, or relative
density distribution. Here, distributions of reference Poisson's
ratio, reference visco Poisson's ratio, reference density must be
distributions of absolute values, while distributions of other
reference elastic constants, and reference visco elastic contants
may be distributions of relative values.
[0838] As the numerical solution method of the first order partial
differential equations, finite difference method or finite element
method can be utilized. By utilizing the regularized algebraic
equations, if the strain tensor field data is contaminated with
errors (noises), or if the reference medium or reference region is
small, or if the reference position is ill-conditioned, the
following distribution can be stably estimated, i.e., shear modulus
distribution, Poisson's ratio distribution, visco shear modulus
distribution, visco Poisson's ratio, density, etc.
[0839] Referring to FIG. 1 again, next explain is about the means
of data processing 1, i.e., calculation method of shear modulus
distribution, Poisson's ratio distribution, visco shear modulus
distribution, visco Poisson's ratio distribution, delay time
distributions, relaxation time distributions, or density
distribution, etc. When the 3D strain tensor, the 3D strain rate
tensor, the 3D acceleration vector, etc. are measured, for
instance, on the Cartesian coordinate system (x,y,z), the next
simultaneous first order partial equations from (125) to (137") are
dealt with, where the shear modulus distribution .mu., the
Poisson's ratio distribution .nu., the visco shear modulus
distribution .mu.', the visco Poisson's ratio distribution .nu.',
the delay time distributions .tau., the relaxation time
distributions .tau.', the strain tensor field .epsilon., the strain
rate tensor field .epsilon..
[0840] That is, when the 3D strain tensor is measured, and only the
shear modulus distribution .mu. is unknown, the next equations are
dealt with, 130 { ij + ij } ( ln ) , j + { ij + ij } , j = 0 , (
125 ) where = v 1 - 2 v , or ( 125 ' ) { ij + ij } , j + { ij + ij
} , j = 0 , ( 126 ) where = v 1 - 2 v . ( 126 ' )
[0841] When the 3D strain tensor is measured, and the shear modulus
distribution .mu. and the Poisson's ratio distribution .nu. are
unknown, the next equations are dealt with, 131 { ij } , j + { ij }
, j + 2 ij , j + 2 ij , j = 0 , ( 127 ) where = 2 v 1 - 2 v . ( 127
' )
[0842] When the 3D strain tensor and the 3D strain rate tensor are
measured, and the shear modulus distribution .mu. and the visco
shear modulus distribution .mu. are unknown, the next equations are
dealt with, 132 { ij + ij } , j + { ij + ij } , j + { ' ' ij + ij '
} , j ' + { ' ' ij + ij ' } , j ' = 0 , ( 128 ) where = v 1 - 2 v ,
( 128 ' ) ' = v ' 1 - 2 v ' , ( 128 " ) or [ t ' t ( t - ) ( t - )
exp { - ( t - ) ( t - ) ' ( t - ) ' ( t - ) ( t - ) } ' ( ) ij ] ,
j + [ t ' t ( t - ) exp { - ( t - ) ' ( t - ) ( t - ) } ij ' ( ) ]
, j = 0 , ( 128 ''' )
[0843] where t' is initial time. If either the shear modulus
distribution .mu. or the visco shear modulus distribution .mu.' is
given, the next equations can be dealt with,
{.phi..epsilon..sub..alpha..alpha..delta..sub.ij+.epsilon..sub.ij}.mu.={.p-
hi.'.epsilon.'.sub..alpha..alpha..delta..sub.ij+.epsilon.'.sub.ij}.mu.'.
(128"")
[0844] If both the shear modulus distributions and the visco shear
modulus distribution .mu.' are unknown, from this equations, the
relaxation times .mu.'(t)/.mu.(t) can be calculated, and can be
utilized in the above equations (128'").
[0845] When the 3D strain tensor and the 3D strain rate tensor are
measured, and the shear modulus distribution, the Poisson's ratio
distribution .nu., the visco shear modulus distribution .mu.', and
the visco Poisson's ratio distribution .nu.' are unknown, the next
equations are dealt with, 133 { ij } , j + { ij } , j + 2 ij , j +
2 ij , j + { ' ij } , j ' + { ' ij } , j ' + 2 ij ' , j ' + 2 ij ,
j ' ' = 0 , ( 129 ) where = 2 v 1 - 2 v , ( 129 ' ) ' = 2 v ' 1 - 2
v ' ' , ( 129 " ) or [ t ' t ( t - ) exp { - ( t - ) ' ( t - ) ( t
- ) } ' ( ) ij ] , j + 2 [ t ' t ( t - ) exp { - ( t - ) ' ( t - )
( t - ) } ij ' ( ) ] , j = 0 , ( 129 ''' )
[0846] where t' is initial time. Either both the shear modulus
distribution .mu. and the visco shear modulus distribution .mu.' or
both the Poisson's ratio distribution .nu. and visco Poisson's
ratio distribution .nu.' are given, the next equations can be dealt
with,
[0847] From this equations, the relaxation time .mu.'(t)/.mu.(t)
can always be calculated. Then if either the shear modulus
distribution .mu. or the visco shear modulus distribution .mu.' is
given, the obtained shear modulus distribution .mu. and the
obtained visco shear modulus distribution .mu.' can be utilized in
the above equations (129'"). Otherwise, if either the Poisson's
ratio distribution .nu. or the visco Poisson's ratio distribution
.nu.' is given, the obtained Poisson's ratio distribution .nu., the
obtained visco Poisson's ratio distribution .nu.', and the obtained
relaxation time .lambda.' (t)/.lambda.(t) can be utilized in the
above equations (129'").
[0848] Equations (128'"), (128""), (129'"), and (129"") can be
dealt with when the target is fluid such as water, secretions,
blood, etc., or tissue includes the fluid much. The equations can
also be dealt with after first temporal partial differentiation or
partial integration. Theoretically, the elastic constant
distributions and visco elastic constant distributions need to be
invariant from the initial time t' to time t.
[0849] When the 2D strain tensor, the 2D strain rate tensor, etc.
are measured, the simultaneous first order partial equations from
(125) to (129"") [i, j=1,2] or the next simultaneous first order
partial equations from (130) to (134"") [i, j=1,2] are dealt with.
The equations from (125) to (129"") [i, j=1,2] are dealt with
approximately under plane strain condition, while the equations
from (130) to (134"") [i, j=1,2] are dealt with approximately under
plane stress condition.
[0850] When the 2D strain tensor is measured, and only the shear
modulus distribution .mu. is unknown, the next equations are dealt
with, 134 { ij + ij } ( ln ) , j + { ij + ij } , j = 0 , ( 130 )
where = v 1 - v , ( 130 ' ) or { ij + ij } , j + { ij + ij } , j =
0 , ( 131 ) where = v 1 - v . ( 131 ' )
[0851] When the 2D strain tensor is measured, and the shear modulus
distribution .mu. and the Poisson's ratio distribution .nu. are
unknown, the next equations are dealt with, 135 { ij } , j + { ij }
, j + ij , j + ij , j = 0 , ( 132 ) where = v 1 - v . ( 132 ' )
[0852] When the 2D strain tensor and the 2D strain rate tensor are
measured, and the shear modulus distribution .mu. and the visco
shear modulus distribution .mu.' are unknown, the next equations
are dealt with, 136 { ij + ij } , j + { ij + ij } , j + { ' ' ij +
ij ' } , j ' + { ' ' ij + ij ' } , j ' = 0 , ( 133 ) where = v 1 -
v , ( 133 ' ) ' = v ' 1 - v ' , ( 133 " ) or [ t ' t ( t - ) ( t -
) exp { - ( t - ) ( t - ) ' ( t - ) ' ( t - ) ( t - ) } ' ( ) ij ]
, j + [ t ' t ( t - ) exp { - ( t - ) ' ( t - ) ( t - ) } ij ' ( )
] , j = 0 , ( 133 ''' )
[0853] where t' is initial time. If either the shear modulus
distributions or the visco shear modulus distribution .mu.' is
given, the next equations can be dealt with,
{.phi..epsilon..sub..alpha..alpha..delta..sub.ij+.epsilon..sub.ij}.mu.={.p-
hi.'.epsilon.'.sub..alpha..alpha..delta..sub.ij.epsilon.'.sub.ij}.mu.'.
(133"")
[0854] If both the shear modulus distributions and the visco shear
modulus distribution .mu.' are unknown, from this equations, the
relaxation time .mu.'(t)/.mu.(t) can be calculated, and can be
utilized in the above equations (133'").
[0855] When the 2D strain tensor and the 2D strain rate tensor are
measured, and the shear modulus distribution .mu., the Poisson's
ratio distribution .nu., the visco shear modulus distribution
.mu.', and the visco Poisson's ratio distribution .nu.' are
unknown, the next equations are dealt with, 137 { ij } , j + { ij }
, j + ij , j + ij , j + { ' ij } , j ' + { ' ij } , j ' + ij ' , j
' + ij , j ' ' = 0 , ( 134 ) where = v 1 - v , ( 134 ' ) ' = v ' 1
- v ' ' , ( 134 " ) or [ t ' t ( t - ) exp { - ( t - ) ' ( t - ) (
t - ) } ' ( ) ij ] , j + [ t ' t ( t - ) exp { - ( t - ) ' ( t - )
( t - ) } ij ' ( ) ] , j = 0 , ( 134 ''' )
[0856] where t' is initial time. Either both the shear modulus
distribution .mu. and the visco shear modulus distribution .mu.' or
both the Poisson' s ratio distribution .nu. and visco Poisson's
ratio distribution .nu.' are given, the next equations can be dealt
with,
.gamma..epsilon..sub..alpha..alpha..delta..sub.ij.epsilon..sub.ij.mu.=.gam-
ma.'.epsilon.'.sub..alpha..alpha..delta..sub.ij+.epsilon.'.sub.ij.mu.'.
(134"")
[0857] From this equations, the relaxation timer .mu.'(t)/.mu.(t)
can always be calculated. Then if either the shear modulus
distribution .mu. or the visco shear modulus distribution .mu.'is
given, the obtained shear modulus distribution .mu. and the
obtained visco shear modulus distribution .mu. can be utilized in
the above equations (134'"). Otherwise, if either the Poisson's
ratio distribution .nu. or the visco Poisson's ratio distribution
.nu.' is given, the obtained Poisson's ratio distribution .nu., the
obtained visco Poisson's ratio distribution .nu.', and the obtained
relaxation time .gamma.'(t)/.gamma.(t) can be utilized in the above
equations (134'").
[0858] Equations (133'"), (133""), (134'"), and (134"") can be
dealt with when the target is fluid such as water, secretions,
blood, etc., or tissue includes the fluid much. The equations can
also be dealt with after first temporal partial differentiation or
partial integration. Theoretically, the elastic constant
distributions and visco elastic constant distributions need to be
invariant from the initial time t' to time t.
[0859] When the 1D strain tensor, the 1D strain rate tensor, etc.
are measured, the simultaneous first order partial equations from
(135) to (137") are dealt with.
[0860] When the 1D strain tensor is measured, and only the shear
modulus distribution .mu. is unknown, the next equations are dealt
with,
.epsilon..sub.11(ln.mu.).sub.,1+.epsilon..sub.11,1=0, (135)
or
.epsilon..sub.11.mu..sub.,1+.epsilon..sub.11,1.mu.=0. (136)
[0861] When the 1D strain tensor and the 1D strain rate tensor are
measured, and the shear modulus distribution .mu. and the visco
shear modulus distribution .mu.' are unknown, the next equations
are dealt with,
.epsilon..sub.11.mu..sub.,1+.epsilon..sub.11,1.mu.+.epsilon.'.sub.11.mu.'.-
sub.,1+.epsilon.'.sub.11,1.mu.'=0, (137)
[0862] or 138 [ t ' t ( t - ) exp { - ( t - ) ' ( t - ) ( t - ) }
11 ' ( ) ] , 1 = 0 , ( 137 ' )
[0863] where t' is initial time. If either the shear modulus
distribution .mu. or the visco shear modulus distribution .mu.' is
given, the next equations can be dealt with,
.epsilon..sub.11.mu.=.epsilon.'.sub.11.mu.'. (137")
[0864] If both the shear modulus distributions and the visco shear
modulus distribution .mu.' are unknown, from this equation, the
relaxation time .mu.'(t)/.mu.(t) can be calculated, and can be
utilized in the above equations (137').
[0865] Equations (137') and (137") can be dealt with when the
target is fluid such as water, secretions, blood, etc., or tissue
includes the fluid much. The equations can also be dealt with after
first temporal partial differentiation or partial integration.
Theoretically, the shear modulus distribution and visco shear
modulus distribution need to be invariant from the initial time t'
to time t.
[0866] In the equations (125), (130), (135), changed can be the
signs of the terms not including (ln.mu.),j, and together changed
can be (ln.mu.),j into {ln (1/.mu.)},j, then resultant partial
differential equations can be dealt with for unknown ln (1I/.mu.).
Although regarding with equations (125), (130), (135) unknown ln
cases are explained below, in unknown ln(1/.mu.) cases ln .mu. or
.mu. can be estimated after ln(1/.mu.) or (1/.mu.) are estimated in
similar ways.
[0867] In the equations (126), (131), (136), changed can be the
signs of the terms not including .mu., j, and together changed can
be .mu. into (1/.mu.), then resultant partial differential
equations can be dealt with for unknown (1/.mu.). Although
regarding with equations (126), (131), (136) unknown .mu. cases are
explained below, in unknown (1/.mu.) cases .mu. or ln.mu. can be
estimated after (1/.mu.) or ln(1/.mu.) are estimated in similar
ways.
[0868] These can be effective when the ROI includes extremely high
shear modulus object such as bone, interstitial needle (for biopsy
and treatment), etc.
[0869] When the target is fluid such as water, secretions, blood,
etc., or tissue includes the fluid much, in the equations (125),
(126), (127), (130), (131), (132), (135), (136) the elastic
constants can be changed into the corresponding visco elastic
constants, and the strain tensor can be changed into the strain
rate tensor. Also in this case, in the equations (125), (130),
(135), changed can be the signs of the terms not including
(ln.mu.'),j, and together changed can be (ln.mu.'),j into {ln
(1/.mu.')},j, then resultant partial differential equations can be
dealt with for unknown ln (1/.mu.').Although regarding with
equations (125), (130), (135) unknown ln.mu.' cases are explained
below, in unknown ln(1/.mu.') cases ln.mu.' or .mu.' can be
estimated after ln(1/.mu.') or (1/.mu.') are estimated in similar
ways.
[0870] In the equations (126), (131), (136), changed can be the
signs of the terms not including .mu.', j, and together changed can
be .mu.'into (1/.mu.'), then resultant partial differential
equations can be dealt with for unknown (1/.mu.'). Although
regarding with equations (126), (131), (136) unknown .mu.' cases
are explained below, in unknown (1/.mu.') cases .mu.' or ln.mu.'
can be estimated after (1/.mu.') or ln(1/.mu.') are estimated in
similar ways.
[0871] These can be effective when the ROI includes extremely high
shear modulus object such as bone, interstitial needle (for biopsy
and treatment), etc.
[0872] When elasticity or visco elasticity is anisotropic,
correspondingly derived equations from (125) to (137") can be dealt
with.
[0873] Regarding density distribution .rho., measured acceleration
vector field a is used. Specifically, in equations (126), (128),
(128'"), (131), (132), (133), (133'"), (134), (134'"), (1/2)
.rho.a.sub.i is added to right terms, in equations (127), (129),
(129'") .rho.a.sub.i is added to right terms, and in equations
(136), (137), (137') (1/3) .rho.a.sub.i is added to right term. The
known density distribution is used in the region, and the unknown
density distribution is estimated with the unknown shear modulus
distribution .mu., the unknown Poisson's ratio distribution .nu.,
the unknown visco shear modulus distribution .mu.', and the unknown
visco Poisson's ratio distribution .nu.'. When the target is fluid
such as water, secretions, blood, etc., or tissue includes the
fluid much, in the equations (126), (127), (131), (132), (136) the
elastic constants can be changed into the corresponding visco
elastic constants, and the strain tensor can be changed into the
strain rate tensor. The density can not be handled when partial
differential equations (126), (131), (136) are directly solved for
ln(1/.mu.), (1/.mu.), ln(1/.mu.'), and (1/.mu.').
[0874] Specifically, according to the measured deformation field,
i.e., the strain tensor field, the strain rate tensor field [when
dealing with the density .rho. (below omitted), the acceleration
vector field, the temporal first derivative of the acceleration
vector field, the strain tensor field, the strain rate tensor
field] and/or the accuracy of the measured deformation field, dealt
with all over the 3D SOI 7 are the simultaneous first order partial
differential equations from (125) to (129""), or dealt within the
plural 3D SOIs, plural 2D ROIs, plural 1D ROIs set in the 3D SOI 7
are respectively the simultaneous first order partial differential
equations from (125) to (129""), the simultaneous first order
partial differential equations from (125) to (134""), the first
order partial differential equations from (135) to (137"). When
plural independent deformation fields are measured, according to
the accuracy of the measured deformation fields, freely either of
the equations from (125) to (137") or the plural equations of the
equations from (125) to (137") can be dealt with at each point of
interest. That is, these equations are solved individually or
simultaneously. The plural independent deformation fields can be
generated under the different positions of the mechanical sources
and the reference regions. These 3D SOIs, 2D ROIs, and 1D ROIs can
include same regions in the 3D SOI 7.
[0875] The Poisson's ratio and visco Poisson's ratio can
respectively be approximated from ratios of the principal values of
the strain tensor and strain rate tensor (on 3D measurement, either
of three ratios of the principle values, or three or two mean
values of the ratios). When plural deformation fields are measured,
the Poisson's ratio and the visco Poisson's ratio can respectively
be approximated as the mean value of the calculated ones from the
plural fields. Typical values can also be utilized for the
Poisson's ratio and the visco Poisson's ratio. For instance, the
object is assumed to be incompressible, then the values are
approximated as the value of about 0.5. Particularly, on equations
from (130) to (134""), the object can be assumed to be completely
incompressible, then the values are approximated as 0.5.
[0876] As initial conditions, at least at one reference point, or
at least in properly set one wide reference region, given are
reference shear modulus, reference Poisson's ratio, reference visco
shear modulus, reference visco Poisson's ratio, etc.
[0877] That is, reference shear moduli (absolute or relative
values) are given at least in one reference region {overscore
(.omega.)}.sub..mu.,1(1- =1.about.N.sub..mu.).
ln.mu.(x,y,z)=ln{circumflex over (.mu.)}(x,y,z), {overscore
(.omega.)}.sub..mu.,1.epsilon.(x,y,z) (138)
.mu.(x,y,z)={circumflex over (.mu.)}(x,y,z), {overscore
(.omega.)}.sub..mu.,1.epsilon.(x,y,z) (138')
[0878] That is, reference Poisson's ratios (absolute values) are
given at least in one reference region {overscore
(.omega.)}.sub..nu.,1(1=1.about.- N.sub..nu.).
.nu.(x,y,z)={circumflex over (.nu.)}(x,y,z), {overscore
(.omega.)}.sub..nu.,1.epsilon.(x,y,z) (139)
[0879] That is, reference visco shear moduli (absolute or relative
values) are given at least in one reference region {overscore
(.omega.)}.sub..mu.',1(1=1.about.N.sub..mu.').
.mu.'(x,y,z)={circumflex over (.mu.)}.sup.1(x,y,z), {overscore
(.omega.)}.sub..mu.',1.epsilon.(x,y,z) (140)
[0880] That is, reference visco Poisson's ratios (absolute values)
are given at least in one reference region {overscore
(.omega.)}.sub..nu.',1(- 1=1.about.N.sub..nu.').
.nu.'(x,y,z)={circumflex over (.nu.)}.sup.1(x,y,z), {overscore
(.omega.)}.sub..nu.',1.epsilon.(x,y,z) (141)
[0881] When elasticity or visco elasticity is anisotropic,
correspondingly derived equations from (125) to (137") and
correspondingly derived initial conditions from (138) to (141) can
be dealt with.
[0882] On discrete Cartesian's coordinate (x,y,z).about.(I.DELTA.x,
J.DELTA.y, K.DELTA.z) in ROI 7 finite difference approximation or
finite element method based on the Galerkin's method or the
variational principle is applied to the shear modulus distribution
.mu., the Poisson's ratio distribution .nu., the elastic constant
distribution .phi., the elastic constant distribution .lambda., the
elastic constant distribution .phi., the elastic constant
distribution .gamma., the visco shear modulus distribution .mu.',
the Poisson's ratio distribution .nu.', the visco elastic constant
distribution .phi.', the visco elastic constant distribution
.lambda.', the visco elastic constant distribution .phi.', the
visco elastic constant distribution .gamma.', the displacement
distribution, the strain distribution, and the strain rate
distribution. Then algebraic equations are derived from the first
order partial differential equations and initial conditions, and
usually the algebraic equations are normalized, for instance, by
the root square of the summation of the powers of the spatially
inhomogeneous coefficients (or the distributions) multiplied to the
shear modulus (distribution) .mu., the Poisson's ratio
(distribution) .nu., the elastic constant (distribution) .phi., the
elastic constant (distribution) .lambda., the elastic constant
(distribution) .phi., the elastic constant (distribution) .gamma.,
the visco shear modulus (distribution) .mu.', the Poisson's ratio
(distribution) .nu.', the visco elastic constant (distribution)
.phi.', the visco elastic constant (distribution) .lambda.', the
visco elastic constant (distribution) .phi.', the visco elastic
constant (distribution) .gamma.'. Furthermore, the algebraic
equations can be regularized. Here, elastic constants .lambda. and
.mu. are called as Lame's constants, while visco elastic constants
.lambda.' and .mu.' are called as visco Lame's constants.
[0883] For instance, finite difference method is utilized, the
simultaneous equations are derived.
EGs=e (142)
[0884] s is unknown vector comprised of the unknown shear modulus
distribution .mu., the unknown elastic constant distribution
.lambda., the unknown elastic constant distribution .gamma., the
unknown visco shear modulus distribution .mu.', the unknown visco
elastic constant distribution .lambda.', the unknown visco elastic
constant distribution .gamma.', etc. G is coefficients matrix
comprised of finite approximations of the 3D, 2D or 1D partial
derivatives. E and e are respectively matrix and vector comprised
of strain tensor data, strain rate tensor data, their derivatives,
and given elastic constants, or visco elastic constants.
[0885] Equations (142) is solved by least squares method, where in
order to reduce the noises of the measured strain tensor data and
strain rate tensor data, the strain distribution and the strain
rate distribution are determined as spatially, temporally, or
spatio-temporally low pass filtered ones. However, inverse of EG
amplifies the high frequency noises filled with e. Then, s becomes
unstable. Thus, to stabilize s the regularization method is
applied. Utilizing the regularization parameters .alpha.1 and
.alpha.2 (at least larger than zero), next equation (143) is
minimized with respect to s, where T indicates transpose.
error
(s)=.vertline.e-EGs.vertline..sup.2+.alpha.1.vertline.Ds.vertline..s-
up.2.vertline.+.alpha.2.vertline.D.sup.TDs.vertline..sup.2
(143)
[0886] D and D.sup.TD are respectively 3D, 2D, or 1D gradient and
Laplacian operator of the unknown shear modulus distribution .mu.,
the unknown elastic constant distribution .lambda., the unknown
elastic constant distribution .gamma., the unknown visco shear
modulus distribution .mu.', the unknown visco elastic constant
distribution .lambda.', the unknown visco elastic constant
distribution .gamma.', etc. That is, with respect to each unknown
distribution, the regularization method can be applied over the 3D
SOI, plural 3D SOIs, 2D ROIs, or 1D ROIs. As Ds and DTD are
positive definite, error (s) absolutely has one minimum value. By
minimizing error(s), the next regularized normal equations are
derived.
(G.sup.TE.sup.TEG+.alpha.1D.sup.TD+.alpha.2D.sup.TDD.sup.TD)s=G.sup.TE.sup-
.Te (144)
[0887] Therefore, the solusion is obtained as
s=(G.sup.TE.sup.TEG+.alpha.1D.sup.TD+.alpha.2D.sup.TDD.sup.TD).sup.-1G.sup-
.TE.sup.Te. (145)
[0888] When the finite element method is utilized, in similar ways,
the least squares method and the regularization method are applied
to the derived simultaneous equations. In this case, G is comprised
of basis function of the unknown nodal elastic modulus distribution
and the unknown nodal visco elastic modulus distribution. Moreover,
utilizing the regularization parameter .alpha.0 (at least larger
than zero), .alpha.0.vertline.s.vertline..sup.2 and
.alpha.0.vertline.Gs.vertline..su- p.2 can be added to the equation
(143). Furthermore, instead of .alpha.1.vertline.Ds.vertline..sup.2
and .alpha.2.vertline.D.sup.TDs.vert- line.,
.alpha.1.vertline.DGs.vertline. and
.alpha.2.vertline.D.sup.TDGs.ve- rtline. can also be utilized.
[0889] The regularization parameter of important information is set
relatively large. Thus, the regularization parameter utilized for
each constant depends on deformation measurement accuracy (SNR),
deformation state, configurations of mechanical sources and
reference regions, number of the utilized independent deformation
fields, etc.; then position of the unknown constant, direction of
the partial derivative, etc.
[0890] From the ratio of each elastic constant E with respect to
the corresponding visco elastic constant E',i.e., (E'/E) for
instance, when measured are the shear modulus, the Poisson's ratio,
the Lame constants, etc. and their corresponding visco elastic
modulus, estimated can be the time delay distribution .tau. [case
when visco elastic modulus is determined from (128), (129), (133),
(134), (136), or(137)] or relaxation time distribution .tau.' [case
when visco elastic modulus is determined from (128'"), (128""),
(129'"), (129""), (133'"), (133""), (134'"), (134""), (136'"),
(136""), (137'), or (137"), or case when visco elastic modulus is
determined from (125), (126), (127), (130), (131), (132), (135), or
(136) where the elastic moduli and strain tensor components are
respectively changed into the corresponding visco elastic mouli and
the strain rate tensor components]. Moreover, from strain tensor
data and elastic moduli data, elastic energy distribution can be
obtained, while from strain rate tensor data and visco elastic
moduli data, consumed energy distribution can be obtained.
[0891] These elastic constants and visco elastic constants can be
temporally changed. Thus, the spatial and temporal simultaneous
equations of the above-described equation can be handled, where
above-described regularization method can spatially and temporally
be applied.
[0892] If the time sequence of the elastic modulus distribution or
the visco elastic modulus distribution is estimated, by spectrum
analysis, the distribution of the frequency variance of the elastic
modulus or the visco elastic modulus can approximately be obtained.
Moreover, if the time sequence of the time delay distribution or
the relaxation time distribution is estimated, by spectrum
analysis, the distribution of the frequency variance of the time
delay or the relaxation time can approximately be obtained. When
estimating the distributions of the frequency variances of these
elastic modulus, visco elastic modulus, time delay, relaxation
time, the deformation field is measured with changing the frequency
of the mechanical source, or with utilizing broadband mechanical
source. Furthermore, at each time, from strain tensor data and
elastic moduli data, elastic energy distribution can be obtained,
while from strain rate tensor data and visco elastic moduli data,
consumed energy distribution can be obtained.
[0893] When solving by the iterative method such as the conjugate
gradient method equations from (143) to (145) derived from
equations from (125) to (137") for each unknown elastic modulus
distribution and each unknown visco elastic modulus distribution,
as explained below, if necessary, newly the reference regions are
set in the ROI in addition to the pre-described reference regions,
and properly initial values of the estimates are set in the unknown
region. In general, each initial value is set based on the a priori
information such as homogeneity and inhomogeneity. Thus,
calculation amount can be reduced.
[0894] Regarding with elasticity distribution, for instance, on 1D
measurement based on the partial differential equation (135) or
(136), by analytically solving these equations, the relative shear
modulus of the point x=X with respect to the point x=A can be
estimated from the ratio of the strains .epsilon.(A)/.epsilon.(X)
(Japanese Patent Application Publication JP-7-55775). This is
effective when tissues deforms in x direction. (Moreover, regarding
with visco elasticity distribution, for instance, on 1D measurement
based on the partial differential equation (135) or (136), by
analytically solving these equations, the relative visco shear
modulus of the point x=X with respect to the point x=A can be
estimated from the ratio of the strain rates
.epsilon.'(A)/.epsilon.'(X). Below, the shear modulus is dealt
with, for instance.)
[0895] However, for instance, in the singular points or the
singular regions where the strain is numerically zero, or the sign
of the strain changes, the shear modulus can be stably estimated
with the above-described regularization method using the absolute
reference values or the relative reference values (reference values
obtained from ratio of the strains in addition to given reference
values.). Otherwise, in the unknown points or the unknown regions
where the absolute strain is less than the positive value A
(threshold), in a similar way, the shear modulus can be stably
estimated using the absolute reference values or the relative
reference values (reference values obtained from ratio of the
strains in addition to given reference values.). In these cases,
the initial values utilized for solving the equations from (143) to
(145) can be determined with various interporation method
(quadrature interporation, cosine interporation, Lagrange's
interporation, spline interporation) such that the values are
spatially continuous with the reference values and the initial
values determined from the a priori information. The threshold A
being dependent on the power or the accuracy (SNR) of the strain
data at each point, the threshold can be spatio-temporally
changeable. The threshold can be set as small value when or where
the SNR of the strain is high, while the threshold can be set as
large value when or where the SNR of the strain is low. Otherwise,
in the unknown points or the unknown regions where the relative
shear modulus values obtained from stain ratio with respect to the
reference values are larger than the relative value B (threshold),
in a similar way, the shear modulus can be stably estimated using
the absolute reference values or the relative reference values
(reference values obtained from ratio of the strains in addition to
given reference values.). Also in this case, the initial values can
be determined with various interporation method such that the
values are spatially continuous with the reference values and the
initial values determined from the a priori information. The
threshold B being dependent on the power or the accuracy (SNR) of
the strain data at each point, the threshold can be
spatio-temporally changeable. The threshold can be set as high
value when or where the SNR of the strain is high, while the
threshold can be set as low value when or where the SNR of the
strain is low. The strain distribution data to determine reference
regions is moving-averaged with the spatio-temporally changeable
window. Otherwise, to properly set the reference regions (values)
and the initial values, the initial values can be calculated with
various interporation method (including linear interporation), and
freely the reference values and initial values can be
spatio-temporally low pass filtered. However, given .mu.(A) is
unchangeable. Also on other equations, the reference regions should
be widely set, in a similar way, the initial values, the singular
points or the singular regions, the unknown points or the unknown
regions can be dealt with. The method to set reference regions
explained here can also be adopted when the direct method is
utilized.
[0896] When solving equations from (143) to (145) derived from
equations from (125) to (137") by the iterative method for each
unknown elastic modulus distribution and each unknown visco elastic
modulus distribution, by properly setting initial values of the
estimates, calculation amount can be reduced. For instance, when
solving equation (135) or (136) for unknown shear modulus
distribution, the initial values can be determined from the
above-described strain ratio. In the above-described singular
points, the singular regions, the points or regions where the
absolute strain is less than the positive value A (threshold), or
the points or regions where the relative shear modulus values
obtained from stain ratio with respect to the reference values are
larger than the relative value B (threshold), the initial values
can be determined with various interporation method (quadrature
interporation, cosine interporation, Lagrange's interporation,
spline interporation) such that the values are spatially continuous
with the reference values and the initial values (the initial
values determined from the a priori information or strain ratio).
Otherwise, in the above-described singular points, the singular
regions, the points or regions where the absolute strain is less
than the positive value A (threshold), or the points or regions
where the relative shear modulus values obtained from stain ratio
with respect to the reference values are larger than the relative
value B (threshold), to properly set the initial values, the
initial values can be calculated with various interporation method
(including linear interporation) from the reference values and the
initial values (the initial values determined from the a priori
information or strain ratio), and freely the reference values and
the initial values can be spatio-temporally low pass filtered.
However, given L (A) is unchangeable. These thresholds being
dependent on the power or the accuracy (SNR) of the strain data at
each point, these thresholds can be spatio-temporally changeable.
The thresholds A and B can respectively be set as small and high
values when or where the SNR of the strain is high, while the
thresholds A and B can respectively be set as large and low values
when or where the SNR of the strain is low. Regarding with other
elastic modulus distributions or other visco elastic modulus
distributions, in a similar way, the initial values can be dealt
with.
[0897] Regarding with some elastic moduli and visco elastic moduli,
as above-explained the reference regions (reference values) and the
initial estimates are set and utilized, and simultaneously other
elastic moduli and visco elastic moduli can be dealt with.
[0898] During iterative estimation, if elastic modulus, visco
elastic modulus, time delay, relaxation time, density are estimated
as the values out of the a priori known ranges, they are corrected
such that they are satisfied with the a priori data. For instance,
the (visco) elastic moduli are positive values. The (visco)
Poisson's ratio is less than 0.5. Then, for instance, if the
(visco) elastic moduli are estimated as negative values, they are
corrected as positive values but nearly equals to zero. If the
(visco) Poisson's ratio are estimated to be larger than 0.5, they
are corrected to be smaller than 0.5 but nearly equals to 0.5. If
plane stress condition is assumed, the (visco) Poisson's ratio can
be corrected to be 0.5.
[0899] On the 1D or 2D measurement of the elastic constants such as
the shear modulus, the Poisson's ratio, etc., and visco elastic
constants such as the visco shear modulus, the visco Poisson's
ratio, etc., they are estimated to be smaller than the original
values when the point of interest gets far from the mechanical
source. In this case, the same shape model having homogeneous
elastic modulus and visco elastic modulus and the same mechanical
source model are utilized, the analytically or numerically
estimated strain data and strain rate data can be utilized to
correct the measured strain data and strain rate data. Otherwise,
on this model analytically or numerically estimated stress data can
be utilized to correct measured elastic modulus distribution and
visco elastic modulus distribution. Otherwise, on this model the
elastic modulus and visco elastic modulus are estimated from the
analytically or numerically estimated strain data and strain rate
data, and the estimates can be utilized to correct measured elastic
modulus distribution and visco elastic modulus distribution.
[0900] The temporal absolute change of the elastic constants such
as the shear modulus, the Poisson's ratio, etc., visco elastic
constants such as the visco shear modulus, the visco Poisson's
ratio, etc., time delay, relaxation time can be obtained as the
difference of the estimated absolute values. The temporal relative
change of the elastic constants, visco elastic constants, time
delay, relaxation time can be obtained as the ratio of the
estimated absolute or relative values, or regarding with the
elastic constants or the visco elastic constants, the temporal
relative change can be obtained as the difference of the estimated
logarithms of them. In this way, on signal processing regarding
with the elastic constants or the visco elastic constants, the
logarithm can be utilized.
[0901] When iteratively solving the equations from (143) to (145),
the initial estimate can be obtained from previous time estimate;
reducing the calculation amount. During iterative estimation, if
elastic modulus, visco elastic modulus, time delay, relaxation
time, density are estimated as the values out of the a priori known
ranges, they are corrected such that they are satisfied with the a
priori data. For instance, the (visco) elastic moduli are positive
values. The (visco) Poisson's ratio is less than 0.5. Then, for
instance, if the (visco) elastic moduli are estimated as negative
values, they are corrected as positive values but nearly equals to
zero. If the (visco) Poisson's ratio are estimated to be larger
than 0.5, they are corrected to be smaller than 0.5 but nearly
equals to 0.5. If plane stress condition is assumed, the (visco)
Poisson's ratio can be corrected to be 0.5.
[0902] The above-explained regularization parameter can be set
larger value when the point of interest gets far from the reference
region along dominant tissue deformation direction.
[0903] On equations from (125) to (137"), the spectrum of the
unknown elastic constants and unknown visco elastic constants are
handled, where the regularization method can be applied not only in
the above-described spatio-temporal directions but also in the
frequency direction.
[0904] For instance, in the 1D ROI (x axis), when measurement
target are frequency variance (spectrum component distribution and
phase distribution) of the sequence of shear modulus distribution
.mu.(x,t) and the sequence of visco shear modulus distribution
.mu.'(x,t), the discrete sequence .mu.(x,j) [j=t/.DELTA.t
(=0.about.n)] of .mu.(x,t) can be expressed as 139 ( x , j ) = 1 n
+ 1 l = 0 n [ ( x , l ) exp ( j ( x , l ) ) ] [ cos ( 2 l fj t ) +
j sin ( 2 l fj t ) ]
[0905] where .mu.(x,l) and .theta..sub..mu.(x,l) are respectively
the spectrum component of the frequency l and the phase of the
frequency l. j expresses imaginary unit. l (=0.about.n) is the
discrete frequency coordinate (f=l.DELTA.f).
[0906] The discrete sequence .mu.'(x,j) [j=t/.DELTA.t (=0.about.n)]
of .mu.'(x,t) can be expressed as 140 ' ( x , j ) = 1 n + 1 l = 0 n
[ ' ( x , l ) exp ( j ' ( x , l ) ) ] [ cos ( 2 l fj t ) + j sin (
2 l fj t ) ]
[0907] where .mu.'(x,l) and .theta..sub..mu.'(x,l) are respectively
the spectrum component of the frequency l and the phase of the
frequency l.
[0908] Then, the first order differential equation (137) can be
expressed as 141 l = 0 n [ ( xx ( x , j ) x ( x , l ) + x xx ( x ,
j ) ( x , l ) ) exp ( j ( x , l ) ) + ( xx ' ( x , j ) x ' ( x , l
) + x xx ' ( x , j ) ' ( x , l ) ) exp ( j ' ( x , l ) ) ] { cos (
2 l fj t ) + j sin ( 2 l fj t ) } = 0 ( 146 )
[0909] Thus, with respect to each frequency l, the following
simultaneous first order differential equations hold. 142 ( xx ( x
, j ) x ( x , 1 ) + x xx ( x , j ) ( x , 1 ) ) cos ( ( x , 1 ) ) +
( xx ' ( x , j ) x ' ( x , 1 ) + x xx ' ( x , j ) ' ( x , 1 ) ) cos
( ' ( x , 1 ) ) = 0 ( 146 ' ) ( xx ( x , j ) x ( x , 1 ) + x xx ( x
, j ) ( x , 1 ) ) sin ( ( x , 1 ) ) + ( xx ' ( x , j ) x ' ( x , 1
) + x xx ' ( x , j ) ' ( x , 1 ) ) sin ( ' ( x , 1 ) ) = 0 ( 146 ''
)
[0910] The simultaneous differential equations (146') and (146")
can be finite difference approximated or finite element
approximated in the same way where the equation (137) is dealt with
at each time j (=0.about.n).
[0911] By substituting the known nodal distribution of the real
components and imaginary components of the spectrum of the
frequency l(=0.about.n) of the elastic constant and visco elastic
constant [.mu.(I,l) cos .theta..sub..mu.(I,l), .mu.(I,l) sin
.theta..sub..mu.(I,l), .mu.'(I,l) cos .theta..sub..mu.'(I,l),
.mu.'(I,l) sin .theta..sub..mu.'(I,l)], at each time j
(j=0.about.n), simultaneous equations (142) are derived each for
real components .mu.(I,l) cos .theta..sub..mu.(I,l) and .mu.'(I,l)
cos .theta..sub..mu.'(I,l), and imaginary components .mu.(I,l) sin
.theta..sub..mu.(I,l) and .mu.'(I,l) sin
.theta..sub..mu.'(I,l).
[0912] In this way, on equations from (125) to (137"), the
simultaneous equations are derived respectively for real components
of the spectrum of the elastic constants and visco elastic
constants, and imaginary components of the spectrum of the elastic
constants and visco elastic constants. When respective simultaneous
equations are regularized, as above-explained, usually, for
instance, the derived algebraic equations are normalized by the
root square of the summation of the powers of the spatially
inhomogeneous coefficient distributions multiplied to the unknown
distributions.
[0913] (A) Two equations derived on each sequence i (=1.about.M),
each time j (=0.about.n), each frequency l (=0.about.n), are
respectively solved for of the frequency l real component
distributions and imaginary component distributions of the spectrum
of the unknown parameters.
[0914] (B) Respective two equations derived from different sequence
i (=1.about.M), different time j (=0.about.n),are simultaneously
set for of the frequency l real component distributions and
imaginary component distributions of the spectrum of the unknown
parameters, and solved.
[0915] (C) Respective two equations derived from different sequence
i (=1.about.M), different time j (=0.about.n), are simultaneously
set for of the frequency l real component distributions and
imaginary component distributions of the spectrum of the unknown
parameters, and by spatial regularization stably solved.
[0916] (D) Respective two equations derived from different sequence
i (=1.about.M), different time j (=0.about.n),are simultaneously
set for of the frequency l real component distributions and
imaginary component distributions of the spectrum of the unknown
parameters, and by temporal regularization stably solved.
[0917] (E) Respective two equations derived on each sequence i
(=1.about.M), each time j (=0.about.n), each frequency l
(=0.about.n), are simultaneously set for real component
distributions and imaginary component distributions of the spectrum
of the unknown parameters, and by frequency regularization stably
solved. Spatial, and temporal regularization can also
simultaneously be performed.
[0918] As above-explained, by one of (A), (B), (C), (D), (E), the
frequency variances of the unknown elastic constants and visco
elastic constants can be obtained.
[0919] The sequences of the nodal elastic constant distributions
and nodal visco elastic constant distributions can be obtained by
inverse Fourier's transform of the spectrums. For instance, the
sequence of the nodal shear modulus distribution is 143 ( I , j ) =
1 n + 1 j = 0 n [ ( I , 1 ) exp ( j ( I , j ) ) ] [ cos ( 2 1 fj t
) + j sin ( 2 1 fj t ) ] ,
[0920] from which the sequence of the shear modulus distribution
.mu.(x,t) can be obtained.
[0921] On also equations from (125) to (137"), the sequences of the
nodal elastic constant distributions and nodal visco elastic
constant distributions can be obtained by inverse Fourier's
transform of the spectrums.
[0922] The deformation fields are measured with changing the
frequency of the mechanical source, or by utilizing broadband
mechanical source.
[0923] When instantaneous frequency of the deformation data can be
measured, the frequency l can be dealt with as the instantaneous
frequency.
[0924] Fourier's transform can be applied not only for time
direction but also spatial direction.
[0925] On equations (126), (127), (128), (129), (131), (132)
(133),(134), (136), (137) and (128'"), (128""),(129'"), (129""),
(133'"), (133""), (134'"), (134""), (137'), (137") in order to deal
with frequency variances of the sequences of the elastic constants
and visco elastic constants, (126), (127), (128), (128""), (129),
(129""), (131), (132), (133), (133""), (134), (134""), (136),
(137), (137") can be approximated utilizing convolute integration
as like (128'"), (129'"), (133'"), (134'"), (137'). For instance,
equation (137) can be approximated as 144 [ t ' t ( t - ) 11 ' ( )
] , 1 + [ t ' t ' ( t - ) 11 " ( ) ] , 1 = 0 , ( 137 ''' )
[0926] where t' is initial time, .epsilon.".sub.11(t) is first
order derivative of the strain rate .epsilon.'.sub.11(t).
[0927] As like on (128'"), (129'"), (133'"), (134'"); (137'),
regularization can be performed temporally and spatially.
[0928] After Fourier's transform, regularization can also be
performed spatial direction, time direction, and in frequency
domain. For instance, equation (137'"):
[N(I,l)E'.sub.11(I,l)].sub.,1+[N'(I,l)E'.sub.11(I,l)].sub.,1=0,
[0929] where E'.sub.11(I,l) is Fourier's transform of the strain
rate .epsilon.'.sub.11 (I,j), and E".sub.11(I,l) is Fourier's
transform of the first order derivative of the strain rate
.epsilon.".sub.11(I,j). From Fourier's transform E.sub.11(I,l) and
E'.sub.11(I,l) respective of the strain E (I,j) and the strain rate
.epsilon..sub.11(I,j), E'.sub.11(I,l) and E".sub.11(I,l) can be
obtained as 145 E 11 ' ( x , 1 ) = ( j2 1 f ) E 11 ( x , 1 ) E 11 "
( x , 1 ) = ( j2 1 f ) E 11 ' ( x , 1 ) = ( j2 1 f ) 2 E 11 ( x , 1
) . ( 137 '''' )
[0930] When dealing with density, density can also be obtained
through regularization.
[0931] In order to determine the unknown elastic constant
distributions, the unknown visco elastic constant distributions,
the unknown density distribution, equations from (125) to (137")
can also be solved utilizing elastic constant data, visco elastic
constant data, density data obtained on equations from (125) to
(137"), and other deformation data.
[0932] Next, utilizing the flowchart of FIG. 26, explained is
measurement procedure of the elastic constant distributions such as
the shear modulus, the Poisson's ratio, etc., the visco elastic
constant distributions such as the visco shear modulus, the visco
Poisson's ratio, etc., time delay distributions, relaxation time
distributions, and the density distributions. At first, reference
regions are properly set for the unknown elastic constants, the
unknown visco elastic constants, the unknown density (S11).
Otherwise, as the reference region, reference points are set in the
ROI 7. Reference point has known elastic constants, known visco
elastic constants, or known density. Otherwise, the reference point
has the reference unity value, or other finite reference
values.
[0933] To obtain high accuracy the elastic constants, the visco
elastic constants, and the density, the reference regions should be
set such that they should widely cross the direction of the
dominant tissue deformation. The reference region has known elastic
constant distributions, known visco elastic constant distributions,
known density distribution, or a priori assumed distributions. When
measuring the absolute elastic constant distributions, the absolute
visco elastic constant distributions, the absolute density
distribution, the given reference values must be absolute
values.
[0934] Occasionally, by assuming the stress distribution in the
reference region, from measured strain values the reference elastic
constant is determined. (For instance, by assuming the stress
distribution to be constant, from strain ratio the reference
elastic constant can be determined.). Moreover, by assuming the
stress distribution in the reference region, from strain rate
values the reference visco elastic constant is determined. (For
instance, by assuming the stress distribution to be constant, from
strain rate ratio the reference visco elastic constant can be
determined.).
[0935] When there exist neither reference point nor reference
regions, if reference medium can directly be contacted to object,
the deformations (strain tensor field, strain rate tensor field,
acceleration vector field) are measured over the ROI including the
reference (S12). In this case, the shear modulus value of the
reference should be large compared with that of the target. The
reference medium should be put between the mechanical source 8 and
the ROI.
[0936] As the object is deformed in 3D space, 3D reconstruction
should be carried out. However, when estimating in the superficial
tissues the elastic constants, the visco elastic constants, and the
density, 1D reconstruction method [from (135) to (137")] is useful
since utilized can be accurately measured strain data, strain rate
data, and acceleration data in beam direction. In contrast, when
estimating in the deeply situated tissues the elastic constants,
the visco elastic constants, and the density, multi-dimensional
reconstruction method is useful since the freedom of configurations
can be high of mechanical sources and reference regions
(mediums)
[0937] Specifically, on 2D reconstruction, when 2D strain
distribution approximately occurs, equations from (125) to (129"")
can be utilized. Alternatively, when 2D stress distribution
approximately occurs, equations from (130) to (134"") can be
utilized. To measure independent deformation fields (strain tensor
fields, strain rate tensor fields, acceleration vector fields), the
position of the mechanical source 8 is changed. Since the
measurement accuracy of the strains, strain rates, acceleration
vectors rely on their magnitudes, to measure the elastic constants,
the visco elastic constants, the density with uniform accuracy over
the ROI, the position of the mechanical source 8 should be
variously changed. This measurement accuracy has the relationship
of trade off between the measurement time and the cost. As already
described, when the object is spontaneously deformed due to
mechanical sources 8' and 8", the mechanical source 8 may not be
needed.
[0938] The measurement controller 3 controls the positions of the
object 6 and the displacement (strain) sensor 5, and the
measurement controller 3 inputs the position information and the
echo signals into the storage 2. At the data processor 1, measured
strain data, strain rate data, acceleration data are filtered to
reduce noises (S13), by which spatially smoothed coefficients E and
e are obtained (S14). Subsequently, the elastic constant
distributions, the visco elastic constant distributions, the
density distribution are obtained from the normal equations (144)
(S15). Thus, measurement results are, at each time, displacement
vector distribution, strain tensor distribution, gradient
distribution of the strain tensor, strain tensor rate distribution,
gradient distribution of the strain rate tensor, elastic constant
distributions such as shear modulus, Poisson's ratio, Lame
constants, etc., visco elastic constant distributions such as visco
shear modulus, visco Poisson's ratio, visco Lame constants, etc.,
time delay distributions or relaxation time distributions relating
these elastic constants and visco elastic constants, density
distribution, gradient distributions of these results, Laplacian
distributions of these results, temporal first derivatives of these
results, temporal second derivatives of these results. To store
time series of these measurement results, these measurement results
(output of data processor 1) are input into the storage 2. To
display in real time these measurement results on CRT (color or
gray), the output of data processor 1 can be input into display
equipment. Freeze image can also be displayed. When displaying
these measurement results, each measurement result can be truncated
by respectively set upper value or lower value. When displaying
elastic constant distributions or visco elastic constant
distributions, the inversion can be also displayed. Moreover,
direct current can be added to the measurement results, or
subtracted from the measurement results. When displaying strain
tensor distribution, to make the sign of the strain invariant, the
direct current can be added (the brightness should be assigned such
that the strain image has correlation with the elastic constant
image). Furthermore, each measurement result can also be displayed
in log scaled.
[0939] Measurement results are, at each time, displacement vector
distribution, strain tensor distribution, gradient distribution of
the strain tensor, strain tensor rate distribution, gradient
distribution of the strain rate tensor, elastic constants such as
shear modulus, Poisson's ratio, Lame constants, etc., visco elastic
constants such as visco shear modulus, visco Poisson's ratio, visco
Lame constants, etc., time delay distributions or relaxation time
distributions relating these elastic constants and visco elastic
constants, density distribution, gradient distributions of these
results, Laplacian distributions of these results, temporal first
derivatives of these results, temporal second derivatives of these
results, time series of these results, time series of relative
(ratio) changes of these results or absolute (substraction) changes
of these results, frequency variance distributions of these
results, elastic energy at each time or accumulations, consumed
energy at each time or accumulations, time series of elastic energy
at each time or accumulations, consumed energy at each time or
accumulations, time series of relative (ratio) changes of these
energy or absolute (substraction) changes of these energy. When
there exist no stain data point or no strain data region, the
elastic constants etc. are interporated or extraporated from
measured ones. The results can be stored at storage 2, and can be
displayed.
[0940] These results can be obtained through spatial filtering of
the absolute elastic constant distributions, absolute visco elastic
constant distributions, absolute time delay distributions, absolute
relaxation time distributions, absolute density distributions, or
relative these distributions obtained from the normal equations
(144). Otherwise, obtained these results can be spatially filtered.
Otherwise, these results can be obtained through temporal or
spatial or temporal-spatial filtering of the elastic constant
distributions, visco elastic constant distributions, time delay
distributions, relaxation time distributions, density
distributions. Otherwise, obtained these results can be temporally
or spatially or temporal-spatially filtered. These results can be
stored at storage 2, and can be displayed. The spatial filter, the
temporal filter, and the temporal-spatial filter are high pass
type, band pass type, low pass type. These filters can be freely
utilized at data processor 1.
[0941] From equations (125) to (137"), the unknown elastic constant
distribution, the unknown visco elastic constant distribution, the
unknown density distribution can be obtained from the measured
elastic constant distribution, visco elastic constant distribution,
density distribution using another deformation field data, or
obtained from their typical value distributions.
[0942] By utilizing the ultrasonic diagnosis equipment together,
the spatial variations of the bulk modulus and the density can be
measured together, and can be displayed together. In this case,
utilized together are the data processor 1, the data storage 2, the
measurement controller 3, displacement (strain) sensor 5,
transmitting/output controller 5' etc (FIG. 1). By utilizing the
magnetic nuclear imaging equipment together, the atomic density
distribution can be measured together, and can be displayed
together.
[0943] As above-described (FIG. 1), using the displacement (strain)
sensor, remotely measured can be strain tensor field, strain rate
tensor field, acceleration vector field. By solving by finite
difference method or finite element method the first order partial
differential equations whose coefficients are derived from the
measured data, estimated can be the absolute elastic constant
distributions, the relative elastic constant distributions with
respect to the reference elastic constants, the absolute visco
elastic constant distributions, the relative visco elastic constant
distributions with respect to reference elastic constants, the
absolute density distribution, the relative density distribution
with respect to reference density.
[0944] By using the regularized algebraic equations, the errors
(measurement noises) of the measured strain data, strain rate data,
acceleration data can be coped with. Moreover, ill-conditioned
reference regions can also be coped with.
[0945] Elasticity and visco-elasticity constants measurement
apparatus, related to this conduct form is useful to monitor
treatment effectiveness of radiation therapy since degeneration and
change of temperature has high correlation with the change of
elastic constants such as shear modulus, Poisson's ratio, Lame
constants, etc., visco elastic constants such as visco shear
modulus, visco Poisson's ratio, visco Lame constants, etc., delay
times or relaxation times relating these elastic constants and
visco elastic constants, and density.
[0946] On conduct form of FIG. 1, as an example, the ultrasound
transducer is utilized as the displacement (strain) sensor 5 to
measure strain tensor, strain rate tensor, acceleration vector. On
the present invention, however, strain tensor, strain rate tensor,
acceleration vector can be measured by signal processing of the
magnetic nuclear resonance signals, and from these deformation data
measured can be elastic constants such as shear modulus, Poisson's
ratio, Lame constants, etc., visco elastic constants such as visco
shear modulus, visco Poisson's ratio, visco Lame constants, etc.,
delay times or relaxation times relating these elastic constants
and visco elastic constants, and density.
[0947] The next explanation is the treatment apparatus related to
one of conduct forms of the present invention. This treatment
apparatus uses for ultrasound therapy the above-explained
measurement technique of displacement vector field/strain tensor
field, and measurement technique of elastic constants, visco
elastic constants, and density.
[0948] The aim of the measurement of the followings is to
quantitatively examine statically or dynamically the objects,
substances, materials, living things, etc., i.e., displacement
vector distribution, the strain tensor distribution, the strain
rate tensor distribution, the acceleration vector distribution, the
velocity vector distribution, elastic constant distributions, visco
elastic constant distributions. For instance, on human soft
tissues, the tissues can be differentiated by extracorporeally
applying pressures or low frequency vibrations, namely, by focusing
on the change of the elasticity due to growth of lesion or change
of pathological state. Instead of the pressures and the vibrations
spontaneous hear motion or pulse can also be utilized to measure
tissue deformation, and tissues can be differentiated from the
values and distributions of tissue elastic constants and visco
elastic constants. Blood velocity can also be observed.
[0949] FIG. 27 shows the global structure of the treatment
apparatus related to this conduct form. On therapy field, lesions
can be treated by applying high intensity ultrasound, laser,
electromagnetic RF wave, electromagnetic micro wave, or by
cryotherapy. On these low invasive treatments, degeneration occurs,
composition rate in weight changes, and temperature changes. For
instance, on living tissues, protein degenerates, and tissue
coagulates. The degeneration, change of composition rate, and
change of temperature occur together with changes of elastic
constants such as shear modulus, Poisson's ratio, etc., visco
elastic constants such as visco shear modulus, visco Poisson's
ratio, etc., delay times or relaxation times relating these elastic
constants and visco elastic constants, or density.
[0950] Thus, by measuring lesion's absolute or relative shear
modulus, absolute or relative Poisson's ratio, absolute or relative
visco shear modulus, absolute or relative visco Poisson's ratio,
absolute or relative delay times or absolute or relative relaxation
times, or absolute or relative density, etc., and by observing
these time courses or these frequency variances, effectiveness of
the treatments can be low invasively monitored. Based on conversion
data for each tissue obtained from theories, simulations, and
measurements, changes of measured shear modulus, Poisson's ratio,
visco shear modulus, visco Poisson's ratio, delay time, relaxation
time, density, strain, strain rate, can be converted into consumed
electric energy, time course of electric energy, temperature, or
time course of temperature. From the measured consumed electric
energy, the time course of electric energy, the temperature, or the
time course of temperature, effectiveness of the treatment can be
confirmed.
[0951] The consumed electric energy, the time course of electric
energy can also be measured by using electric power meter and
tissue physical parameters (tissue electric impedance, mechanical
impedance, etc.). The temperature, or the time course of
temperature can also be measured by using usual temperature
monitoring method, thermo coupler, etc. By measuring these spatial
distributions, not only effectiveness of treatment can be
monitored, but also safety and reliability can be obtained. These
monitoring data can be utilized for dynamic electronic digital
control or mechanical control of beam focus position, treatment
interval, ultrasound beam power, ultrasound beam strength, transmit
term, transmit interval, beam shape (apodization), etc. Thus, these
monitoring data can be utilized to improve the efficiency of the
treatment.
[0952] FIG. 27 shows the treatment apparatus which transmits high
intensity ultrasounds to lesion. The treatment apparatus can be
equipped with ultrasound diagnosis equipment and elasticity and
visco-elasticity constants measurement apparatus. As shown in FIG.
27, the treatment probe 11 possesses the ultrasound transducer 12
and treatment transducer 13 (treatment transducer can also serve as
the ultrasound transducer 12), supporter 14. As utilized on the
ultrasonic diagnosis equipment (for instance, convex type
transducer), the ultrasound transducer 12 arrays plural
oscillators. The treatment transducer 13 also arrays plural
oscillators. In the figure, concavity type treatment probe 11 is
shown. The supporter 14 can be held by hand or position controller
4, by which the position of the treatment probe 11 can be
controlled.
[0953] To the treatment transducer 13, the ultrasound pulse
generated at the treatment pulse generator 21 is provided through
the treatment wave delay circuit 22 and amplifier 23. That is, at
the treatment wave delay circuit 22 the delay time of the transmit
ultrasound pulse is controlled for each oscillator, by which the
focus position of the synthesized ultrasound beam is
controllable.
[0954] To the oscillators of the ultrasound transducer 12, the
ultrasound pulses generated at the ultrasound pulse generator 31
are provided through the transmit and receive separator 34 after
being focused at the transmit delay circuit and being amplified at
the amplifier 33. The echo signals received at the oscillators of
the ultrasound transducer 12 are amplified at the amplifier 35
after passing through the transmit and receive separator, and the
phases of the echo signals are matched at the phase matcher 36. The
outputs of the phase matcher 36 are used to reconstruct image at
the signal processor 37, and the image data is converted to
diagnosis image at the DSC (digital scan converter) 38, and the
diagnosis image is displayed at the monitor 39. Known ultrasound
diagnosis equipment can be used for this diagnosis equipment.
[0955] The elastic constants and visco elastic constants
measurement part 40 related to this conduct form can measure shear
modulus, Poisson's ratio, visco shear modulus, visco Poisson's
ratio, density, delay times or relaxation times relating these
elastic constants and visco elastic constants, density, etc., using
the echo signals output from phase matcher 36. The measured data
and calculated results are stored at the data storage equipped with
40.
[0956] Commands of the controller 41 control the treatment pulse
generator 21, the treatment wave delay circuit 22, the ultrasound
pulse generator 31, the transmit delay circuit 32, the phase
matcher 36, the signal processor 37, the DSC 38, and the elastic
constants and visco elastic constants measurement part 40. The
operator can input commands and conditions from the operation part
42 into the controller 41, by which the operator can set various
operation conditions and treatment conditions. The signal processor
37, elastic constants and visco elastic constants measurement part
40, operation part 42, controller 41 are comprised of
computers.
[0957] Next explanation is how this like ultrasound treatment
equipment is utilized. The treatment probe 11 is contacted onto
body surface, and is supported such that the ROI include the target
lesion. Occasionally, by using water tank, the treatment probe 11
is supported without contacting onto the body surface. At first, to
image the lesion part, the command to start imaging is input from
the operation part 42, by which as the response the controller 41
outputs the commands to the ultrasound pulse generator 31 and the
transmit delay circuit 32. Then, the ultrasound beam is transmitted
from the ultrasound transducer 12. This ultrasound beam scans the
ROI. The echo signals are received at the oscillators of the
ultrasound transducer, the phases of the echo signals are matched
at the phase matcher 36. The outputs of the phase matcher 36 are
used to reconstruct image at the signal processor 37, and the image
data is converted to diagnosis 2D image at the DSC (digital scan
converter) 38, and the diagnosis image is displayed at the monitor
39. Thus, during observing the images and diagnosing tissues, when
the lesion part can be detected, treatment is carried out.
[0958] That is, when the lesion is detected, the treatment probe is
held at the present position. From the image memorized at the DSC
38, the controller 41 obtains the delay time to provide the drive
pulse to each oscillator of the treatment transducer. Then, the
controller outputs the obtained time delays into the treatment wave
delay circuit 22, by which the lesion part is focused. The strength
of the ultrasound beam can be controlled. The lesion part is
heated. The lesion part degenerates. The treatment can also be
carried out by observing 3D ultrasound image. Controlled of
treatment ultrasound beam can be beam focus position, treatment
interval, ultrasound beam power, ultrasound beam strength, transmit
term, transmit interval, beam shape (apodization), etc.
[0959] Next explanation is the procedure of treatment and
measurement of shear modulus, Poisson's ratio, visco shear modulus,
visco Poisson's ratio, time delay, relaxation time density, etc.
for monitoring the treatment effectiveness. Flowchart of FIG. 28 is
referred to. At first, before the treatment, measured in the ROI
are shear modulus distribution .mu.(x,y,z), Poisson's ratio
.nu.(x,y,z), visco shear modulus .mu.'(x,y,z), visco Poisson's
ratio .nu.'(x,y,z), delay time .tau.(x,y,z), relaxation time
.tau.'(x,y,z), density .rho.(x,y,z) (S21). Command is sent from the
operator part 42 to the controller 41, after which the ultrasounds
are transmitted from the ultrasound transducer 12. Subsequently,
the controller 41 sends command to the elastic constants and visco
elastic constants measurement part 40, by which using echo signals
output from the phase matcher 36 the strain tensor field or strain
rate tensor field are measured. From the measured strain tensor
field or strain rate tensor field, calculated are shear modulus
distribution .mu.(x,y,z), Poisson's ratio .nu.(x,y,z), visco shear
modulus .mu.'(x,y,z), visco Poisson's ratio .nu.'(x,y,z), delay
time .tau.(x,y,z), relaxation time .tau.'(x,y,z), density
.rho.(x,y,z), etc.
[0960] Next, if the lesion part is confirmed, the treatment process
counter I is initialized (I=0) (S22). The starting position of the
treatment and the initial strength of the treatment ultrasound are
set (S23), and the treatment is started (S24). At every treatment,
measured are shear modulus distribution .mu.(x,y,z), Poisson's
ratio distribution .nu.(x,y,z), visco shear modulus distribution
.mu.'(x,y,z), visco Poisson's ratio distribution .nu.'(x,y,z),
delay time distribution .tau.(x,y,z), relaxation time distribution
.tau.'(x,y,z), density distribution .rho.(x,y,z), etc. (S25). The
measured elastic constants, visco elastic constants, delay times,
relaxation times can be absolute values or spatially relative
values or temporally relative values. Then, to confirm the
effectiveness of the treatment, comparison can be carried out
between shear modulus value .mu.(x,y,z),Poisson's ratio value
.nu.(x,y,z), visco shear modulus value .mu.'(x,y,z) visco Poisson's
ratio value .nu.'(x,y,z), etc., and their respective thresholds TH1
(softened case) and TH2 (hardened case), etc. (S26). Moreover,
comparison can be carried out between delay time value
.tau.(x,y,z), relaxation time value .tau.'(x,y,z), density value
.rho.(x,y,z) and their respective thresholds. The thresholds TH1,
TH2, etc., can be set from the information of the tissue properties
etc. The thresholds TH1, TH2, etc. are the functions of the time t,
the position (x,y,z), ultrasound parameters such as shooting
counter etc., degeneration information, etc. The thresholds can be
set before the treatment, or can be updated during the treatment.
If desired effectiveness cannot be confirmed, the ultrasound
strength is controlled to be higher (S27), after which the
treatment is carried out again (S24). If the desired effectiveness
can be confirmed, it is judged if the treatments of all the
positions are finished (S28). If the treatments of all the
positions are not finished yet, the treatment position is changed
(S29), and the treatment is carried out again (S24).
[0961] If the treatments of all the positions are finished, the
treated part is naturally or compulsively cooled down (S30). After
the treatment, measured are shear modulus distribution .mu.(x,y,z),
Poisson's ratio distribution .nu.(x,y,z), visco shear modulus
distribution .mu.'(x,y,z), visco Poisson's ratio distribution
.nu.'(x,y,z), delay time distribution .tau.(x,y,z), relaxation time
distribution .tau.'(x,y,z), density distribution .rho.(x,y,z), etc.
(S31). It is judged if desired effectiveness can be obtained at all
the positions (S32). If the desired effectiveness can not be
confirmed at all the positions, till the effectiveness can be
confirmed, the treated part is cooled down (from S30 to S32). If
the desired effectiveness is confirmed at all the positions, it is
judged if this treatment process is finished (S33). When the
treatment process is not finished, the treatment process counter I
is incremented, and steps from S23 to S33 are iteratively carried
out. The maximum number of the treatment process can be set. The
treatment position can be set in order from deep position or
peripheral position, or the treatment position can be set where the
treatment effectiveness is not confirmed.
[0962] As described above, using the treatment apparatus of FIG.
27, during the ultrasound treatment, we can observe the treatment
effectiveness in real time and then we can properly carry out the
treatment. Moreover, by confirming the treatment effectiveness in
real time, the ultrasound strength, the shoot number, etc. can be
controlled.
[0963] The treatment apparatus of FIG. 27 can also be used for the
other treatments such as laser treatment, electromagnetic RF wave
treatment, electromagnetic micro wave treatment, or cryotherapy. In
this case, the low invasive treatment modalities are substituted
for the treatment probe 11, the treatment pulse generator 21, the
treatment wave delay circuit 22, the amplifier 23.
[0964] As the ultrasound transducer 12, for instance, utilized can
be 2D array aperture type applicator, 1D array aperture type
applicator, concavity type applicator. For instance, when carrying
out cryotherapy or radio therapeutics (applying high intensity
focus ultrasound, laser, electromagnetic RF wave, micro wave, etc.)
on living things or the in vitro tissues through skin, mouth,
vagina, anus, opened body, body surface, monitored can be
degeneration, change of composition rate in weight, and change of
temperature. Measured shear modulus, Poisson's ratio, visco shear
modulus, visco Poisson's ratio delay time, relaxation time,
density, etc., can be utilized as index to dynamically control beam
position (focus), treatment interval, beam power, beam strength,
transmit term, transmit interval, beam shape (apodization),
etc.
[0965] Before, during, after the treatment, the followings can be
displayed on monitor 39 as static or motion or time course
(difference) image, the values of arbitrary points, the time course
(graph), etc., i.e., not only elastic constant distribution or
visco elastic constant but also displacement vector distribution,
displacement vector component distributions, strain tensor
component distributions, strain gradient component distributions,
strain rate tensor component distributions, strain rate gradient
component, acceleration vector component distributions, or velocity
vector component distributions, etc.
[0966] Moreover, by utilizing ultrasound diagnosis apparatus
together, spatial variations of bulk modulus and density of tissues
can be measured and displayed in real-time. On the ultrasound
image, as measurement results, superimposed and displayed can be
static or motion or time course (difference) images of the
displacement vector distribution, displacement vector component
distributions, strain tensor component distributions, strain
gradient component distributions, strain rate tensor component
distributions, strain rate gradient component, acceleration vector
component distributions, velocity vector component distributions,
etc.
[0967] Particularly when the applicator has an arrayed aperture,
beam focus position, treatment interval, beam power, beam strength,
transmit term, transmit interval, beam shape (apodization), etc.
are electronically digital controlled, while when the applicator
has a concavity aperture, the focus position is mechanically
controlled. The flowchart of FIG. 28 can be applied to the control
program, for instance. That is, to control beam focus position,
treatment interval, beam power, beam strength, transmit term,
transmit interval, beam shape (apodization), etc, utilized can be
absolute or relative shear modulus distribution, absolute or
relative Poisson's ratio distribution, absolute or relative visco
shear modulus distribution, absolute or relative visco Poisson's
ratio distribution, absolute or relative delay time distributions,
absolute or relative relaxation time distributions, absolute or
relative density distribution, temporally absolute or relative
changes of these elastic constants, visco elastic constants, delay
times, relaxation times, density, etc. measured before, during,
after transmitting the energies.
[0968] The above-explained measurement technique of displacement
vector field, strain tensor field, etc., and measurement technique
of elastic constants, visco elastic constants, density, etc., can
be utilized together with interstitial needle, catheter, etc. when
carrying out cryotherapy or radio therapeutics (applying high
intensity focus ultrasound, laser, electromagnetic RF wave, micro
wave, etc.) or when non-destructive examining living things or
substances or materials (cases included during producing or
growing.).
[0969] For instance, on interstitial cryotherapy, interstitial
radio therapeutics (applying high intensity focus ultrasound,
laser, electromagnetic RF wave, micro wave, etc. utilizing needles
and plate, only needles, mono needle, etc), etc., the followings
can also be displayed on monitor before, during, after the
treatment as static or motion or time course (difference image)
image, the values of arbitrary points, the time course
(graph),etc., i.e., not only elastic constant distribution or visco
elastic constant but also displacement vector distribution,
displacement vector component distributions, strain tensor
component distributions, strain gradient component distributions,
strain rate tensor component distributions, strain rate gradient
component, acceleration vector component distributions, or velocity
vector component distributions, etc. Moreover, by utilizing
ultrasound diagnosis apparatus together, spatial variations of bulk
modulus and density of tissues can also be measured and displayed
in real-time. On the ultrasound image, as measurement results,
superimposed and displayed can also be static or motion or time
course (difference) images of the displacement vector distribution,
displacement vector component distributions, strain tensor
component distributions, strain gradient component distributions,
strain rate tensor component distributions, strain rate gradient
component, acceleration vector component distributions, velocity
vector component distributions, etc. The followings can be
displayed in vector style as well, i.e., the displacement vector
distribution, acceleration vector, velocity vector.
[0970] To obtain safety when carrying out treatment, by setting the
upper values and lower values of shear modulus, Poisson's ratio,
visco shear modulus, visco Poisson's ratio, delay times, relaxation
times, density, etc., and by setting the upper values of absolute
or relative changes of these, beam position (focus), treatment
interval, beam power, beam strength, transmit term, transmit
interval, beam shape (apodization) etc. should be controlled such
that these physical parameter values do not change more than
necessary.
[0971] The treatment effectiveness can also be evaluated by
measuring temperature and temporal change of temperature as
above-explained from strain (tensor) distribution, strain rate
(tensor) distribution, shear modulus distribution, Poisson's ratio
distribution, visco shear modulus distribution, visco Poisson's
ratio distribution, density distribution, temporal changes of
these, etc. measured before, during, after transmitting the
energies. In this case, to obtain safety, by setting the upper
values of temperature or change of temperature, beam position
(focus), treatment interval, beam power, beam strength, transmit
term, transmit interval, beam shape (apodization), etc. should be
controlled such that the temperature do not heighten more than
necessary. These can also be controlled utilizing shear modulus
value .mu., Poisson's ratio value .nu., visco shear modulus value,
visco Poisson's ratio value, density value, delay time values,
relaxation time values, strain values, strain rate values, etc.
converted from the upper values. Temperature and change of
temperature can also be measured utilizing the conventional
temperature measurement method or thermo coupler.
[0972] In cases where no mechanical source exists, or mechanical
sources are not utilized, degeneration, change of composition rate
in weight, and change of temperature can also be detected from
strain (tensor) distribution, strain rate (tensor) distribution,
shear modulus distribution, Poisson's ratio distribution, visco
shear modulus distribution, visco Poisson's ratio distribution,
density distribution, temporal changes of these, etc. measured
before, during, after transmitting the energies. Directly the
expansion and shrink can also be detected when strain (tensor)
distribution or strain rate (tensor) distribution are measured.
[0973] The elasticity and visco-elasticity constants measurement
apparatus of the present invention can be utilized to monitor
degeneration, change of composition rate in weight, change of
temperature due to injection of medicine, putting of medicine,
giving of medicine. To control amount the medicine, term, interval,
position, etc., utilized can be absolute or relative shear modulus
distribution, absolute or relative Poisson's ratio distribution,
absolute or relative visco shear modulus distribution, absolute or
relative visco Poisson's ratio distribution, absolute or relative
delay time distributions, absolute or relative relaxation time
distributions, absolute or relative density distribution,
temporally absolute or relative changes of these elastic constants,
visco elastic constants, delay times, relaxation times, density,
etc. measured before, during, after the treatment. Anticancer drug
can be utilized as the medicine.
[0974] That is, to monitor the treatment effectiveness (including
change of temperature) of anticancer drug and to control the
treatment, the followings can also be displayed on monitor before,
during, after the treatment as static or motion or time course
(difference) image, the values of arbitrary points, the time course
(graph), etc., i.e., not only elastic constant distribution or
visco elastic constant but also displacement vector distribution,
displacement vector component distributions, strain tensor
component distributions, strain gradient component distributions,
strain rate tensor component distributions, strain rate gradient
component, acceleration vector component distributions, or velocity
vector component distributions, etc. Moreover, by utilizing
ultrasound diagnosis apparatus together, spatial variations of bulk
modulus and density of tissues can also be measured and displayed
in real-time. On the ultrasound image, as measurement results,
superimposed and displayed can also be static or motion or time
course (difference) images of the displacement vector distribution,
displacement vector component distributions, strain tensor
component distributions, strain gradient component distributions,
strain rate tensor component distributions, strain rate gradient
component, acceleration vector component distributions, velocity
vector component distributions, etc. The followings can be
displayed in vector style as well, i.e., the displacement vector
distribution, acceleration vector, velocity vector. In cases where
no mechanical source exists, or mechanical sources are not
utilized, degeneration, expansion or shrink, and change of
temperature, etc. can also be detected from displacement vector,
strain (tensor) distribution, strain rate (tensor) distribution,
etc.
[0975] The elastic constants, visco elastic constants, density,
high order data expressed from elastic constants, visco elastic
constants, density are utilized to obtain non-linear properties of
tissues by linear approximation of non-linear phenomena in
infinitesimal time space or spatial space. Thus, estimated
non-linear elastic constants data, non-linear visco elastic
constants data, high order data expressed from non-linear data can
be utilized for diagnosis and treatment.
[0976] Thus, as explained above, the present invention can realize
accurate measurement in 3D space of interest (SOI) or 2D region of
interest (ROI) or 1D ROI of displacement vector distribution,
strain tensor distribution, the spatio or temporal derivatives of
these, generated due to arbitrary mechanical sources. If the target
naturally deforms, elastic constant or visco elastic constant can
be estimated in the SOI or ROI without disturbing the deformation
field from measured deformation data. Moreover, even if there exist
another mechanical sources and uncontrollable mechanical sources in
the object, for instance, the elastic constant and visco elastic
constant measurement apparatus can be utilized, which is applicable
for diagnosing the part of interest in the object and for
monitoring the treatment effectiveness. Furthermore, low-invasive
treatment apparatus can be realized, which is equipped with such
elastic constant and visco elastic constant measurement
apparatus.
* * * * *