U.S. patent application number 15/666964 was filed with the patent office on 2017-11-16 for system and method for improved energy series of images using multi-energy ct.
The applicant listed for this patent is Mayo Foundation For Medical Education And Research. Invention is credited to Joel G. Fletcher, Shuai Leng, Cynthia H. McCollough, Charles A. Mistretta, Lifeng Yu.
Application Number | 20170330354 15/666964 |
Document ID | / |
Family ID | 45469830 |
Filed Date | 2017-11-16 |
United States Patent
Application |
20170330354 |
Kind Code |
A1 |
Leng; Shuai ; et
al. |
November 16, 2017 |
SYSTEM AND METHOD FOR IMPROVED ENERGY SERIES OF IMAGES USING
MULTI-ENERGY CT
Abstract
A method for creating an energy series of images acquired using
a multi-energy computed tomography (CT) imaging system having a
plurality of energy bins includes acquiring, with the multi-energy
CT imaging system, a series of energy data sets, where each energy
data set is associated with at least one of the energy bins. The
method includes producing a conglomerate image using at least a
plurality of the energy data sets and, using the conglomerate
image, reconstructing an energy series of images, each image in the
energy series of images corresponding to at least one of the energy
data sets.
Inventors: |
Leng; Shuai; (Rochester,
MN) ; McCollough; Cynthia H.; (Byron, MN) ;
Yu; Lifeng; (Byron, MN) ; Fletcher; Joel G.;
(Oronoco, MN) ; Mistretta; Charles A.; (Madison,
WI) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Mayo Foundation For Medical Education And Research |
Rochester |
MN |
US |
|
|
Family ID: |
45469830 |
Appl. No.: |
15/666964 |
Filed: |
August 2, 2017 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
14944666 |
Nov 18, 2015 |
9754387 |
|
|
15666964 |
|
|
|
|
13809800 |
Jan 11, 2013 |
9208585 |
|
|
PCT/US2011/044390 |
Jul 18, 2011 |
|
|
|
14944666 |
|
|
|
|
61365191 |
Jul 16, 2010 |
|
|
|
Current U.S.
Class: |
1/1 |
Current CPC
Class: |
A61B 6/032 20130101;
A61B 6/482 20130101; G01N 2223/1016 20130101; A61B 6/4241 20130101;
G06T 5/20 20130101; G06T 11/003 20130101; G01N 23/046 20130101;
G01N 2223/423 20130101; A61B 6/50 20130101; G01N 2223/419 20130101;
A61B 6/5205 20130101 |
International
Class: |
G06T 11/00 20060101
G06T011/00; G01N 23/04 20060101 G01N023/04; A61B 6/00 20060101
A61B006/00; A61B 6/00 20060101 A61B006/00; A61B 6/03 20060101
A61B006/03; A61B 6/00 20060101 A61B006/00; G06T 5/20 20060101
G06T005/20; A61B 6/00 20060101 A61B006/00 |
Claims
1. A method for creating a series of images acquired using a
multi-energy computed tomography (CT) imaging system having a
plurality of energy bins, the method comprising the steps of: (a)
acquiring a series of energy data sets, each energy data set
associated with at least one of the energy bins; (b) producing a
conglomerate data set using at least a plurality of the energy data
sets; and (c) using the conglomerate data set, generating at least
one of an enhanced material-specific image and an enhanced energy
series of images, each image corresponding to at least one of the
energy data sets.
2. The method of claim 1 wherein the conglomerate data set includes
data from each energy data set in the series of energy data
sets.
3. The method of claim 1 wherein step (b) includes producing a
HYPR-based composite image by combining corresponding views across
each of the series of energy data sets acquired in step (a) and
step (c) includes reconstructing the energy series of images by
normalizing the energy data set with information derived from the
composite image and by multiplying the normalized result with the
composite image.
4. The method of claim 1 wherein the series of energy data sets is
acquired using at least one of a photon counting and energy
discriminating CT detector.
5. The method of claim 1 wherein step (c) includes at least one of
reconstructing material decomposition images using the series of
energy data sets and generating the enhanced material-specific
image using the conglomerate data set.
6. The method of claim 1 further comprising weighting each of the
energy-selective data sets.
7. A method for creating an energy series of images acquired using
a computed tomography (CT) imaging system, the method comprising
the steps of: (a) acquiring a series of energy-selective data sets,
each energy-selective data set associated with energy bin; (b)
producing a conglomerate data set from the energy-selective data
sets including data associated with at least a plurality of the
energy bins; (c) weighting each of the energy-selective data sets
using the conglomerate data set; and (d) reconstructing an enhanced
energy series of images, where each image in the enhanced energy
series of images corresponds to at least one of the energy data
sets.
8. The method of claim 7 wherein step (b) includes producing the
conglomerate data set from all data in the energy-selective data
sets.
9. The method of claim 7 wherein step (c) includes filtering the
series of energy-selective data sets and the conglomerate data
set.
10. The method of claim 9 wherein the filtering includes applying a
low-pass filter kernel.
11. The method of claim 7 wherein step (c) includes forming a
weighting data set as the ratio of a given data set in the series
of energy-selective data sets to the conglomerate data set.
12. The method of claim 7 wherein step (d) includes performing a
HYPR-based reconstruction using the conglomerate data set and the
series of energy selective data sets.
13. The method of claim 7 wherein step (d) includes applying a
HYPR-LR-based processing mathematically expressed as: I HE = I E K
I C K I C ; ##EQU00004## where I.sub.HE represents an image in the
enhanced energy series of images, I.sub.E represents individual
energy data sets in the series of energy-selective data sets,
I.sub.C represents the conglomerate image data set, K represents a
filter kernel, and {circumflex over (x)} represents a convolution
process.
14. The method of claim 7 wherein step (d) includes applying a
HYPR-LR-based processing to form the enhanced energy series of
images as the multiplication of a weighting used in step (c) and
the conglomerate image data set.
15. The method of claim 7 wherein step (a) includes acquiring the
series of energy-selective data sets using a multi-energy CT
imaging system.
16. A computed tomography (CT) imaging system comprising: an x-ray
source configured to emit x-rays toward an object to be imaged; a
detector configured to receive x-rays that are attenuated by the
object; a data acquisition system (DAS) connected to the detector
to receive an indication of received x-rays; a computer system
coupled to the DAS to receive the indication of the received x-rays
and programmed to: segregate the indication of the received x-rays
into a series of energy data sets based on an energy level
associated with received x-rays; produce a conglomerate image data
set using data from at least a plurality of the energy data sets;
and reconstruct at least one of an enhanced material-specific image
and an enhanced energy series of images, each image in the at least
one of the enhanced material-specific image and enhanced energy
series of images corresponding to at least one of the energy data
sets, using the series of energy data sets and the conglomerate
image data set.
17. The CT imaging system of claim 16 wherein the detector includes
at least one of a photon counting and energy discriminating CT
detector.
18. The CT imaging system of claim 16 wherein the conglomerate
image data set is formed using all data in the plurality of the
energy data sets.
19. The CT imaging system of claim 16 wherein the computer is
further programmed to form a weighting image data set as the ratio
of a given data set in the plurality of the energy data sets to the
conglomerate image data set.
20. The CT imaging system of claim 16 wherein the computer is
further programmed to performing a HYPR-based reconstruction to
reconstruct the enhanced energy series of images.
21. The CT imaging system of claim 16 wherein the computer is
further programmed to perform a HYPR-LR-based processing to
reconstruct the enhanced energy series of images as a
multiplication of a weighting and the conglomerate image data
set.
22. A method for creating a series of images acquired using a
multi-energy imaging system having a plurality of energy bins, the
method comprising the steps of: (a) acquiring a series of
energy-selective data sets, each energy-selective data set
associated with at least one of the energy bins of the imaging
system; (b) producing a conglomerate data set using at least a
plurality of the energy-selective data sets; and (c) using the
conglomerate data set, generating at least one of an enhanced
material-specific image and an enhanced energy series of
images.
23. The method of claim 22 wherein the conglomerate data set
includes data from each energy data set in the series of energy
data sets.
24. The method of claim 22 wherein step (b) includes producing a
HYPR-based composite image by combining corresponding views across
each of the series of energy data sets acquired in step (a) and
step (c) includes reconstructing the energy series of images by
normalizing the energy data set with information derived from the
composite image and by multiplying the normalized result with the
composite image.
25. The method of 22 wherein step (c) includes at least one of
reconstructing material decomposition images using the series of
energy data sets and generating the enhanced material-specific
image using the conglomerate data set.
Description
CROSS-REFERENCE TO RELATED APPLICATIONS
[0001] This application is a continuation application of U.S.
patent application Ser. No. 14/944,666 filed Nov. 18, 2015, which
is a continuation application of U.S. patent application Ser. No.
13/809,800 filed Jan. 11, 2013, which is a 371 application of
PCT/US2011/44390 filed Jul. 18, 2011, which claims the benefit of
U.S. Provisional Application No. 61/365,191, filed Jul. 16, 2010,
all of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
[0002] The present invention relates to computed tomography (CT)
imaging and, more particularly, to systems and methods for energy
domain data correction in spectral CT imaging to control noise and
radiation dose.
[0003] In a computed tomography system, an x-ray source projects a
fan or cone shaped beam which is collimated to lie within an X-Y
plane of a Cartesian coordinate system, termed the "imaging plane."
The x-ray beam passes through the object being imaged, such as a
medical patient or other non-medical patient or object, such as in
industrial CT imaging, and impinges upon an array of radiation
detectors. The intensity of the transmitted radiation is dependent
upon the attenuation of the x-ray beam by the object and each
detector produces a separate electrical signal that is a
measurement of the beam attenuation. The attenuation measurements
from all the detectors are acquired separately to produce the
transmission profile at a particular view angle.
[0004] The source and detector array in a conventional CT system
are rotated on a gantry within the imaging plane and around the
object so that the angle at which the x-ray beam intersects the
object constantly changes. A group of x-ray attenuation
measurements from the detector array at a given angle is referred
to as a "view", and a "scan" of the object comprises a set of views
acquired at different angular orientations during one revolution of
the x-ray source and detector. In a 2D scan, data is processed to
construct an image that corresponds to a two dimensional slice
taken through the object. The prevailing method for reconstructing
an image from 2D data is referred to in the art as the filtered
backprojection technique, however, other image reconstruction
processes are also well known. This process converts the
attenuation measurements from a scan into integers called "CT
numbers" or "Hounsfield units", which are used to control the
brightness of a corresponding pixel on a display.
[0005] The term "generation" is used in CT to describe successively
commercially available types of CT systems utilizing different
modes of scanning motion and x-ray detection. More specifically,
each generation is characterized by a particular geometry of
scanning motion, scanning time, x-ray beam shape, and detector
system.
[0006] The first generation utilized a single linear x-ray beam
("pencil beam") and a single scintillation crystal-photomultiplier
tube detector for each tomographic slice. After a single linear
motion or traversal of the x-ray tube and detector, during which
time 160 separate x-ray attenuation or detector readings are
typically taken, the x-ray tube and detector are rotated through
one degree and another linear scan is performed to acquire another
view. This is repeated typically to acquire 180 views.
[0007] A second generation of CT systems was developed to shorten
the scanning times of first generation systems by gathering the
attenuation data more quickly. In these units, a modified fan beam,
including anywhere from 3-52 individual collimated x-ray beams, and
a number of detectors equal to the number of collimated x-ray beams
are used. Individual beams resemble the single beam of a first
generation scanner; however, a collection of from 3-52 of these
beams contiguous to one another allows multiple adjacent regions of
tissue to be examined simultaneously. The configuration of these
contiguous regions of tissue resembles a fan, with the thickness of
the fan material determined by the collimation of the beam and in
turn determining the slice thickness. Because of the angular
difference of each beam relative to the others, several different
angular views through the body slice are being examined
simultaneously. Superimposed on this is a linear translation or
scan of the x-ray tube and detectors through the body slice. Thus,
at the end of a single translational scan, during which time 160
readings may be made by each detector, the total number of readings
obtained is equal to the number of detectors times 160. The
increment of angular rotation between views can be significantly
larger than with a first generation unit, up to as much as
36.degree.. Thus, the number of distinct rotations of the scanning
apparatus can be significantly reduced, with a coincidental
reduction in scanning time. By gathering more data per translation,
fewer translations are needed.
[0008] To obtain even faster scanning times it is necessary to
eliminate the complex translational-rotational motion of the first
two generations. Third generation scanners therefore use a much
wider, "divergent" fan beam. In fact, the angle of the beam may be
wide enough to encompass most or all of an entire patient section
without the need for a linear translation of the x-ray tube and
detectors. As in the first two generations, the detectors, now in
the form of a large array, are rigidly aligned relative to the
x-ray beam, and there are no translational motions at all. The tube
and detector array are synchronously rotated about the patient
through an angle of 180-360.degree.. Thus, there is only one type
of motion, allowing a much faster scanning time to be achieved.
After one rotation, a single tomographic section is obtained.
[0009] Fourth generation scanners also feature a divergent fan beam
similar to the third generation CT system. As before, the x-ray
tube rotates through 360.degree. without having to make any
translational motion. However, unlike in the other scanners, the
detectors are not aligned rigidly relative to the x-ray beam. In
this system only the x-ray tube rotates. A large ring of detectors
are fixed in an outer circle in the scanning plane. The necessity
of rotating only the tube, but not the detectors, allows faster
scan time.
[0010] Beyond these large "generational" distinctions between CT
technology, a number of additional advancements have been made. For
example, dual energy and even dual source CT systems have been
developed. In either case, x-ray dose of different energy levels
are used to acquire two image data sets from which a low energy and
a high energy image may be reconstructed. As will be described, a
wide variety of information can than be determined from the subject
by analyzing the characteristics and variations between the low
energy data set and the high energy data set.
[0011] In addition, photon counting (PC) and energy discriminating
(ED) detector CT systems have the potential to greatly increase the
medical benefits of CT. Unlike the above-described "traditional" CT
detectors, which integrate the charge generated by x-ray photon
interactions in the detector but provide no specific energy
information regarding individual photons, PC detectors record the
energy deposited by each individual photon interacting with the
detector. PC detector system can provide new clinical abilities due
to an ability to differentiate materials such as a contrast agent
in the blood and calcifications that may otherwise be
indistinguishable in traditional CT systems. Also, they can improve
the signal to noise ratio (SNR) by reducing electronic and swank
noise. PC and ED CT systems generally produce less image noise for
the same dose than photon energy integrating detectors and hence
can be more dose efficient than conventional CT systems. Also, they
can improve SNR by assigning optimal, energy dependent weighting
factors to the detected photons and achieve additional SNR
improvements by completely or partially rejecting scattered
photons. Further still, PC detectors allow measurement of
transmitted, energy-resolved spectra from a single exposure at one
tube potential.
[0012] The development of PC detectors for micro-CT and whole-body
CT applications has enabled a new dimension of CT imaging, namely
"spectral CT" or "multi-energy CT." These advances have attracted
considerable attention in the scientific and research communities,
due to the potential for enhanced material characterization
utilizing spectral x-ray information. This lays the groundwork for
many clinical applications, such as detecting new biomarkers, such
as iron in vulnerable plaque, multi-contrast imaging, such as
iodine and barium imaging of the bowel luminal wall and intra-lumen
contents, and exploring intrinsic tissue contrast, cancerous tissue
compared to normal tissue.
[0013] In contrast to conventional CT systems, where photons are
measured and recorded in a single transmission data set, spectral
CT generates multiple data sets, with each data set measuring only
those photons with energies between predefined low and high energy
thresholds. Because the x-ray attenuation of a material depends on
the photon energy, material-specific information is then built into
each energy-specific data set. Measured data from each energy bin
is then reconstructed independently to generate a series of CT
images, each corresponding to a specific energy range. These images
are highly correlated, since the anatomic geometry and physical
density of the object remains unchanged for any time point. Only
the total x-ray attenuation values, that is, CT numbers, differ,
according to the material type and selected photon energy bin. An
attenuation-energy curve can be generated from these multiple image
series, each image corresponding to one energy bin. Since each
material has its own attenuation-energy curve, material
identification/differentiation can then be achieved using
multi-energy CT.
[0014] The appropriate selection of energy bins, for example, the
number of energy bins and width of each energy bin, has a
significant affect on the outcome of spectral imaging. A narrow
energy bin has better energy resolution compared to a wider energy
bin, and hence enables better material
identification/differentiation. For example, FIG. 1 shows a graph
of two energy bins used to separate iron, which is a biomarker for
vulnerable plaques, from calcium. A narrow energy window width of
20 keV was used and the two energy bins were widely separated in
the x-ray spectrum. The dual energy ratio difference, which is an
indicator of material separation capability, or the dual-energy
"contrast" between two materials, using these energy windows were
compared with conventional dual energy CT, in which wider energy
windows, with a low of 0 to 80 kVp and an high of 0 to 140kVp were
used. Significant improvement was observed using the narrow beam
energy windows (20 keV). However, a significant limitation of using
narrow bins is that the number of photons available in each energy
bin is much smaller than the total number of photons detected. For
the scenario in FIG. 1, only a small portion of total photons were
used in each energy-specific image and a large fraction of photons
in between energy bin 1 and 2 were discarded.
[0015] As image noise is proportional to the inverse square root of
available photons, image noise is correspondingly higher using a
narrow energy bin than a wide energy bin. Thus, a critical problem
occurs. Specifically, in order to identify or differentiate
materials using spectral CT, the differences in effective atomic
number or signal must be amplified by: 1) using narrow energy bins
and 2) separating the energy bins as widely as possible. However,
this requirement excludes a large percentage of the detected energy
spectrum from the considered image data. Thus, the resultant
images, in which dual-energy signal is increased, suffer from
increased noise. For narrow energy bins, especially in the lower
energy range, the image noise may be so high as to make it
impossible to detect small differences in material composition,
that is, the signal to noise ratio (SNR) is too low. Further, a
large portion of the dose delivered to the patient is wasted,
creating a difficult dilemma of the clinician balancing between
dose delivered and achieving a desired SNR. Thus, the requirements
for increasing dual-energy signal are in direct conflict with the
requirements for decreasing image noise in the individual energy
images and in any material composition images derived from the
energy specific images.
[0016] Similar observations exist for the selection of total number
of energy bins. For a given x-ray spectrum, a given kVp, more
energy bins provide more measurements of energy dependent
information. With multiple data points available along the
attenuation-energy curve, better curve-fitting, consequently better
material differentiation is achieved. However, more bins also
dictates narrower widths for each bin and hence fewer photons in
each bin. Turning to FIG. 2, a scenario in which 6 energy bins were
used is illustrated. In FIG. 2, 6 separate measurements in the
energy domain corresponding to the 6 energy bins are available.
However, the number of photons in each energy bin is only 1/6 of
the total photons delivered. Accordingly, the noise in each image
is then significantly high.
[0017] Therefore, an intrinsic tradeoff exists in the selection of
energy bins (number, width, and placement) for spectral CT,
resulting in the described tradeoff between energy-specific signal
(material identification/differentiation information) and noise.
This tradeoff limits the clinical applications of spectral CT. For
example, for the differentiation between iron (a biomarker for
plaque vulnerability) from calcium in vascular plaques, narrow
energy bins are generally used due to the very small concentration
iron amidst a typically higher concentration of calcium (i.e. there
is a very weak signal). Due to the small signal size, image noise
must be strictly controlled to allow the detection of iron, and
hence the identification of those plaques more likely to rupture
and cause acute myocardial infarction. Thus, a dilemma is presented
of increasing signal size through appropriate selection of the
energy bins is counterproductive due to the increase in image
noise. Although increased photons (dose) could potentially be used,
increases in patient dose above existing levels will prevent
clinical application due to the heightened concern about ionizing
radiation in medicine and potential long-term effects of such
radiation on patients. An increased dose will also likely require
higher power and cooling requirements on the x-ray tube and
generator, as current coronary CT angiography already uses the
upper limits of tube/generator technology. Addressing this with use
of longer scan times, such as using longer gantry rotation times,
would sacrifice image quality with motion artifact and hence blur
out the small signal that is sought.
[0018] Accordingly, it would be desirable to have a system and
method for creating an energy series of images with reduced noise
and increased signal to noise ratio.
SUMMARY OF THE INVENTION
[0019] The present invention overcomes the aforementioned drawbacks
by providing a system and method for creating an energy series of
images acquired using a multi-energy computed tomography (CT)
imaging system having a plurality of energy bins. Using the
multi-energy CT imaging system, a series of energy data sets is
acquired, where each energy data set is associated with at least
one of the energy bins. A conglomerate image is produced using a
plurality of the energy data sets and, using the conglomerate
image, an energy series of images is reconstructed, where each
image in the energy series of images corresponds to at least one of
the energy data sets. Thus, the present invention seeks to exploit
a correlation of information in the energy domain to reduce image
noise in each energy-specific image, not just in a processed
material-composition image. As such, the present invention improves
material differentiation in spectral CT by allowing selection of
desired energy bins and, in particular, the number of bins, the
width of the bins, and the location of the bins, without paying a
noise penalty.
[0020] In accordance with one aspect of the invention, a method for
creating an energy series of images acquired using a computed
tomography (CT) imaging system is disclosed that includes acquiring
a series of energy-selective data sets, each energy-selective data
set associated with energy bin and producing a conglomerate data
set from the energy-selective data sets including data associated
with at least a plurality of the energy bins. The method also
includes weighting each of the energy-selective data sets using the
conglomerate data set and reconstructing an enhanced energy series
of images, where each image in the enhanced energy series of images
corresponds to at least one of the energy data sets.
[0021] In accordance with another aspect of the invention, a method
for creating an energy series of images acquired using a
multi-energy computed tomography (CT) imaging system having a
plurality of energy bins is disclosed that includes acquiring a
series of energy data sets, each energy data set associated with at
least one of the energy bins. The method also includes producing a
conglomerate data set using at least a plurality of the energy data
sets and using the conglomerate data set, generating at least one
of an enhanced material-specific image and an enhanced energy
series of images, each image corresponding to at least one of the
energy data sets.
[0022] In accordance with still another aspect of the invention, a
computed tomography (CT) imaging system is disclosed that includes
an x-ray source configured to emit x-rays toward an object to be
imaged, a detector configured to receive x-rays that are attenuated
by the object, and a data acquisition system (DAS) connected to the
detector to receive an indication of received x-rays. The system
also includes a computer system coupled to the DAS to receive the
indication of the received x-rays and programmed to segregate the
indication of the received x-rays into a series of energy data sets
based on an energy level associated with received x-rays. The
computer is further programmed to produce a conglomerate data set
using data from at least a plurality of the energy data sets and
reconstruct at least one of an enhanced material-specific image and
an enhanced energy series of images, each image in the at least one
of the enhanced material-specific image and enhanced energy series
of images corresponding to at least one of the energy data sets,
using the series of energy data sets and the conglomerate data
set.
[0023] Various other features of the present invention will be made
apparent from the following detailed description and the
drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
[0024] FIG. 1 is a graph illustrating energy with respect to two
exemplary bins used with photon counting detectors and relative to
a broad brehmsstralung spectrum.
[0025] FIG. 2 is a graph illustrating that a dual-energy signal
between iron and calcium is maximized with use of the two, widely
separated energy bins compared to energy windows currently used
with integrating detector technology (80/140 kV=80 and 140 kVp
beams or 80/140Sn=80 kVp and 140 kVp beams, where the 140 kVp beam
has been filtered with approximately 0.4 mm of tin (Sn)) to
selectively reduce the number of photons below 80 keV, which
increases the dual-energy signal due to the increased separation of
mean beam energy.
[0026] FIG. 3 is pictorial view of a CT imaging system in which the
present invention may be employed.
[0027] FIG. 4 is block schematic diagram of the CT imaging system
of FIG. 3.
[0028] FIG. 5 is a flow chart setting forth exemplary steps of a
method for creating an enhanced series of images in accordance with
the present invention.
[0029] FIG. 6 is a flow chart setting forth exemplary steps of a
method utilizing HYPR-based techniques to create an enhanced series
of images in accordance with the present invention.
[0030] FIG. 7A is a series of images reconstructed without the use
of a conglomerate image spanning all energy bins.
[0031] FIG. 7B is a series of images reconstructed using HYPR-LR to
provide a conglomerate image spanning all energy bins and,
thereafter, reconstruct a series of energy images at 80, 100, 120,
and 140 kVp.
[0032] FIG. 8 is a graph showing CT number and noise measured at
ROIs representing calcium, water, and soft tissue across a number
of reconstruction methods.
DETAILED DESCRIPTION OF THE INVENTION
[0033] With initial reference to FIGS. 3 and 4, a computed
tomography (CT) imaging system 10 includes a gantry 12
representative of at least a "third generation" CT scanner. In the
illustrated example, the gantry 12 has a pair of x-ray sources 13
that each project a fan beam or cone beam of x-rays 14 toward a
detector array 16 on the opposite side of the gantry 12. However,
it is specifically noted that the present invention, while readily
applicable to dual-source, dual-energy CT systems, is also readily
applicable to other multi-energy CT systems and methods, such as
single-source, dual- or multi-energy CT systems and methods. The
detector array 16 is formed by a number of detector elements 18
that together sense the projected x-rays that pass through a
medical patient 15. As will be described, it is contemplated that
the detector array 16 may form part of a so-called
"photon-counting" and/or "energy-discriminating" detector system.
In any case, each detector element 18 produces an electrical signal
that represents the intensity of an impinging x-ray beam or, more
accurately, each photon or bunch of photons, and hence the
attenuation of the beam (or photons) as it passes through the
patient. During a scan to acquire x-ray projection data, the gantry
12 and the components mounted thereon rotate about a center of
rotation 19 located within the patient 15 to acquire attenuation
data.
[0034] The rotation of the gantry and the operation of the x-ray
source(s) 13 are governed by a control mechanism 20 of the CT
system. The control mechanism 20 includes an x-ray controller 22
that provides power and timing signals to the x-ray sources 13 and
a gantry motor controller 23 that controls the rotational speed and
position of the gantry 12. A data acquisition system (DAS) 24 in
the control mechanism 20 samples analog data from detector elements
18 and converts the data to digital signals for subsequent
processing. An image reconstructor 25, receives sampled and
digitized x-ray data from the DAS 24 and performs high speed image
reconstruction. The reconstructed image is applied as an input to a
computer 26 which stores the image in a mass storage device 28.
[0035] The computer 26 also receives commands and scanning
parameters from an operator via console 30 that has a keyboard. An
associated display 32 allows the operator to observe the
reconstructed image and other data from the computer 26. The
operator supplied commands and parameters are used by the computer
26 to provide control signals and information to the DAS 24, the
x-ray controller 22, and the gantry motor controller 23. In
addition, computer 26 operates a table motor controller 34 that
controls a motorized table 36 to position the patient 15 in the
gantry 12.
[0036] In CT scans, image noise is highly correlated to the number
of photons received. Thus, lower noise in a resulting image is
achieved when more x-ray photons are used to create the image. For
this reason, traditional notions of CT imaging focus including all
usable x-ray information. However, spectral CT imaging diverts from
this notion by segregating the information into bins. The present
invention recognizes that spectral CT is, in essence, an imaging
technique in four dimensions. In particular, spectral CT deals with
the three dimensions in space and a unique dimension in energy.
Within this conceptual context, the present invention recognizes
that time-resolved spectral CT includes a 5th dimension, namely a
dimension of time.
[0037] The present invention builds on the above recognition that
spectral CT is, in essence, an imaging technique in four dimensions
(or five dimensions, in the case of a time series of images) and
further recognizes that a high degree of correlation exists between
the energy-specific data sets of spectral CT imaging due to the
fact that the data sets pertain to the same patient anatomy. Using
these recognitions, the present invention exploits this correlation
of information in the energy domain to reduce image noise in each
energy-specific image, not just in a processed material composition
image. That is, although extensive research has been conducted in
the spatial and temporal domains to reduce noise and improve image
quality, generally, limited investigation has been done in energy
domain. Currently, data acquired from each energy bin in the energy
domain is treated independently and CT images at each energy
utilize only data from a single energy bin. As will be described,
the present invention diverts from this traditional notion and
ultimately improves material differentiation in spectral CT by
allowing selection of the optimal or desired energy bins and, in
particular, the number of bins, the width of the bins, and the
location of the bins, without paying a substantial noise
penalty.
[0038] In x-ray CT, image noise is inversely related to the square
root of the total number of photons used to reconstruct the image.
The number of photons associated with each energy bin image in
spectral CT is reduced for the same patient exposure because of
dividing the total number of photons applied to the patient into
multiple energy bins to obtain energy specific information. Noise
level therefore increases compared to conventional CT images that
use all available photons. The degree of noise increase depends on
the number of energy bins and the width of each energy bin. By
utilizing the redundant information in the energy domain, image
noise in spectral CT can be reduced to the level of conventional
CT.
[0039] Thus, the present invention recognizes that, in the case of
multi-energy CT imaging, images reconstructed from all received
photons can be treated as "conventional" CT images that have little
or no material differentiating information (energy-resolved
signal), but which also have the lowest noise and use the full dose
applied to the patient. Comparing these "conventional" images
reconstructed from all received photons and images reconstructed
from each energy bin in a multi-energy acquisition illustrates that
the images are not independent. Rather, the images actually have a
high degree of correlation due to the fact that they measure the
same anatomy. The present invention exploits this correlation to
generate new energy-specific, CT image sets that have dramatically
reduced noise levels, such as is generally achievable with
conventional CT images, yet maintain the CT numbers of individual
energy bin images. These images retain each individual data set's
specific energy signature, while noise can be reduced to as low as
that of the conventional image that is reconstructed directly from
all acquired photons.
[0040] Turning now to FIG. 5, a process for imaging in accordance
with the present invention will be described with respect to
exemplary steps embodied as a flow chart 100. The process begins
with the performance of a CT imaging process, at process block 102,
such as using the above-described CT systems, including PC or ED CT
systems. In this sense, it is contemplated that the CT imaging
process may be a traditional "multi-energy" CT imaging process,
including, for example, the common "dual-energy" and "dual-energy,
dual-source" CT imaging processes. In addition, it is contemplated,
for example, when using the aforementioned PC or ED CT systems,
that the CT process may deviate from traditional "multi-energy" CT
acquisitions, so long as the ability to discriminate and bin the
acquired data based on energy is maintained. To this point, at
process block 104, the acquired CT data is assembled or assigned,
as described above, into bins that serve to divide the acquired CT
data into spectral data sets.
[0041] As noted above, the present invention recognizes that these
spectral data sets are, in essence, multi-dimensional data sets,
where one of the dimensions spans the energy series. For example,
if the acquired CT data is three dimensional (3D) in the sense of
acquiring CT in three spatial dimensions, spectral data sets are
treated as four dimensional (4D) data sets. Similarly, if the
acquired CT data is two dimensional (2D) in the sense of acquiring
CT data in two spatial dimensions or the acquired CT data is 3D in
the sense of acquiring CT data in 3 spatial dimensions and across a
time series, the spectral data sets represent 3D or 5D data sets,
respectively, in the context of the present invention. The present
invention further recognizes that a high degree of correlation
exists between the spectral data sets due to the fact that the data
sets pertain to the same patient anatomy. Using this information,
the present invention, at process block 106, forms a "conglomerate
data set" that includes data spanning the various spectral data
sets assembled at process block 104. For example, data from bins
associated with 20-40 keV, 40-60 keV, 60-80 keV, and the like can
serve as the bins across which the conglomerate data set spans. Put
another way, the conglomerate data set and, any conglomerate image,
is formed from CT from a variety of different energy bins used to
segregate the CT data acquired using the multi-energy imaging
process at process block 102. As will be explained in detail, this
conglomerate data set and/or any associated conglomerate image
formed therefrom may be used to reconstruct an energy series of CT
images at process block 108 that has substantially improved
material differentiation achieved without a substantial noise
penalty incurred using traditional methods.
[0042] In accordance with some implementations, the "conglomerate
data set" or conglomerate image may use a substantial amount or
even most or all x-ray photons received and associated with the
energy bins. By using a conglomerate data set during the
reconstruction of individual energy images from the data associated
with each individual energy bin, the trade-off between bin number
(or width) and image noise is substantially reduced or, for
clinical purposes, effectively eliminated. This provides
previously-unachievable flexibility to choose energy bins based
upon signal optimization so that the best material
identification/differentiation information can be achieved without
the images succumbing to noise. Therefore, the constraints
presented in FIGS. 1 and 2 can be managed in a clinical setting,
without paying the significant penalty of either increased image
noise or increased patient dose.
[0043] A variety of methods are contemplated for creating a
conglomerate data set and reconstructing an energy series of images
using the conglomerate data set or image. For example, two methods
include a HighlY constrained back-PRojection HYPR) processing and
reconstruction and a Prior Image Constrained Compressed Sensing
(PICCS), Non-convex PICCS, and multi-band filtration. Exemplary
HYPR and HYPR-based methods are described in U.S. Pat. No.
7,519,412, which is incorporated herein by reference. Exemplary
PICCS and PICCS-based methods are described in US Patent
Application Publication No. 2009/0161932, which is incorporated
herein by reference.
[0044] In accordance with one aspect of the invention, the
"conglomerate image" may be provided by applying the concepts
creating a "composite image" as described within the context of
HYPR, to use the acquired photon data that spans multiple energy
bins to form the conglomerate data set. HYPR and its modified
version, HYPR-LR, allow the reconstruction of a time or other
series images from highly undersampled data set using a "composite
image" built from multiple time series of images. HYPR-LR and
HYPR-LR-related methods are described in U.S. Patent Publication
No. 2008-0219535, which is incorporated herein by reference. These
HYPR-based concepts, which were first applied to a time series of
images, can be used in accordance with the present invention to
reconstruct an energy series of images formed of separate images,
as will be described.
[0045] It has been demonstrated that the signal to noise ratio
(SNR) of HYPR-reconstructed images are determined by the composite
image instead of the single frame image, which improves the SNR of
images at each time frame. The HYPR technique and general concepts
thereof can be adapted into spectral CT imaging to provide a
"conglomerate data set" or "conglomerate image" to improve SNR of
images at each individual energy bin. In this regard, the
"composite image" of HYPR and the concepts for creation of the
"composite image" may be extended to the above-described energy
series to form a "conglomerate data set" or "conglomerate image" in
the context of the present invention.
[0046] A "conglomerate data set" or "conglomerate image" in the
context of the present invention may be formed using the concept of
a "composite image" in HYPR by using x-ray photons acquired across
the energy spectrum and respective bins and, thereby, a "composite
data set" or "conglomerate data set" can be formed having the SNR
that is independent of the number of energy bins. That is, as the
SNR of HYPR images is determined by the composite image, in the
present invention, images at each energy bin therefore have an SNR
equivalent to that obtained with all x-ray photons. For example,
these conglomerate images can be generated using an averaging or
may be generated by applying different weighting factors to each
energy bin, which may improve image quality. By reconstructing the
individual energy images associated with each energy bin using a
conglomerate image and a HYPR-based reconstruction, the tradeoff
between number of energy bins and image noise in each energy bin is
substantially reduced or eliminated.
[0047] Turning to FIG. 6, a specific example, using the HYPR-LR
concept of a "composite image" and reconstruction, is provided by
way of a flow chart 200. The exemplary process using HYPR-based
techniques begins by performing a CT imaging process. As explained
above, it is contemplated that the CT imaging process may be a
traditional "multi-energy" CT imaging process, including, for
example, the common "dual-energy" and "dual-energy, dual-source" CT
imaging processes. In addition, it is contemplated, for example,
when using the aforementioned PC or ED CT systems, that the CT
process may deviate from traditional "multi-energy" CT
acquisitions, so long as the ability to discriminate and bin the
acquired data based on energy is maintained. To this point, at
process block 204, the acquired CT data is assembled or assigned,
as described above, into bins that serve to divide the acquired CT
data into spectral data sets.
[0048] As previously described with respect to FIG. 5, a
conglomerate data set is then formed from the spectral data sets at
process block 206. However, within this example utilizing
HYPR-based techniques for forming the conglomerate data set and
reconstructing the energy series of images, a conglomerate image
I.sub.C may be produced by averaging data from some or all of the
energy bins. In some cases, all of the data from all of the bins
may be used to ensure that all available photons are included in
the conglomerate image to, thereby, produce the lowest image noise.
At process block 208 a filter operation is then performed on both
individual energy data sets or images I.sub.E and the composite
data set/image I.sub.C. At process block 210, a weighting or
weighting image is obtained as the ratio between individual energy
data sets or images I.sub.E and the composite data set/image
I.sub.C, as filtered. At process block 212, HYPR "processing" or
reconstruction, for example, HYPR-LR processing, can then be used
to form an enhanced spectral image I.sub.HE as the multiplication
of the weighting image and the conglomerate image. Mathematically,
the HYPR-LR algorithm can be expressed as:
I HE = I E K I C K I C ; Eqn . 1 ##EQU00001##
where K is a low-pass filter kernel. A kernel, such as a 733 7
pixel uniform square kernel or other desirable kernel, can be used.
The symbol "{circumflex over (x)}" represents a convolution
process.
[0049] Using error propagation theory, image noise after such
HYPR-LR processing has been derived in MRI images and CT images. It
can be expressed as:
.sigma. I HE 2 .apprxeq. .sigma. I C 2 + .sigma. I E 2 N K + 2
.sigma. I C 2 N K ; Eqn . 2 ##EQU00002##
[0050] where .sigma..sub.I.sub.HE.sup.2 is the noise variance in
the HYPR-LR processed images at energy bin E,
.sigma..sub.I.sub.C.sup.2 is the noise variance in composite image,
.sigma..sub.I.sub.E.sup.2 is the noise variance in individual
energy bin image, and N.sub.K is the number of pixels used in the
filter kernel. It can be observed that the noise variance of
HYPR-LR images is mainly determined by that of the composite images
and this translates to the present invention and the use of HYPR-LR
with a "conglomerate image." Thus, in the present invention, noise
variance is mainly determined by that of the composite images and
only weakly depends on that of individual energy bin image. This
relationship is more obvious if the energy bins are selected in
such a way that similar noise is measured in each individual energy
bin
(.sigma..sub.I.sub.E.sup.2=N.sub.E.times..sigma..sub.I.sub.C.sup.2).
In this scenario, Eqn. 2 can be rewritten as:
.sigma. I HE 2 .apprxeq. .sigma. I C 2 ( 1 + N E + 2 N K ) ; Eq . 3
##EQU00003##
[0051] where N.sub.E is the number of energy bins. In practice,
N.sub.E (number of energy bins) is usually much smaller than
N.sub.K due to physical limitations in the detector hardware.
[0052] Therefore, the noise variance of images
.sigma..sub.I.sub.HE.sup.2 in the enhanced energy series is
expected to be close to that of conglomerate image.
[0053] It is noted that the size of the convolution kernel used for
the HYPR-LR processing has an impact on CT number accuracy and
image noise reduction. As seen from Eqns. (2) and (3), image noise
is reduced as kernel size increases, although the incremental noise
reduction diminishes as the noise level approaches the noise level
of composite image. A very large kernel could affect CT number
accuracy and consequently energy specific information. A 7.times.7
pixel uniform filter kernel has been demonstrated as a reasonable
choice to reduce image noise without affecting CT number accuracy
or spatial resolution. One limitation of the HYPR-LR algorithm is
that it prefers imaging scenarios without substantial motion
between energy-specific images. However, in most spectral CT
systems (e.g. photon-counting, detector-based or dual-source CT
systems), imaging data at different beam energies are acquired
simultaneously. Therefore, motion is of minimal concern. For
systems in which substantial delay is expected between different
energy data acquisitions (e.g. dual energy CT using two separate
scans), motion might be a concern and HYPR-LR techniques should be
adjusted to account for motion.
[0054] To demonstrate this effect, CT numbers and noise variances
can be measured inside three circular ROls placed at regions
representing calcium, water, and soft tissue. The measurement can
be conducted on images reconstructed using both commercial software
and HYPR-LR for comparison. Noise reduction using HYPR-LR is then
calculated, and corresponding dose reduction is estimated based
upon the relationship between radiation dose and image noise.
Images of a semi-anthropomorphic thoracic phantom scanned at 80,
100, 120, and 140 kVp are shown in FIGS. 7A and 7B. Specifically,
FIG. 7A shows images reconstructed without the aid of the present
invention and FIG. 7B shows images processed with the present
invention and HYPR providing the composite image as the
conglomerate image. Significant noise reduction was observed after
HYPR processing for images at each beam energy. Due to the high
image noise in the original image, the water-equivalent rod, which
is labeled in FIG. 7B, is almost not differentiable in the images
of FIG. 7A. On the other hand, in the images of FIG. 7B, the
water-equivalent rod is clearly seen. This distinction between the
images of FIGS. 7A and 7B is due to the use of a conglomerate image
that reduces noise and increases SNR in the resulting images.
[0055] Thus, HYPR-LR processing applied in the context of the
present invention did not alter spatial resolution or
energy-specific CT numbers. The effectiveness of the present
invention readily translates to both photon-counting,
detector-based and integrating, detector-based CT systems using
numerical simulations, phantom, and patient studies. Numerical
simulations demonstrated a 36-76% noise reduction using the present
invention with HYPR-LR processing in the energy domain compared to
standard FBP reconstruction for the case when 6 energy bins were
used. The percent noise reduction changed when different numbers of
energy bins or different bin widths were used.
[0056] Turning now to FIG. 8, the CT number (as mean) and image
noise (as standard variation) of calcium, water, and tissue
equivalent are shown. The energy dependent CT numbers for different
materials makes it possible for material differentiation using
spectral CT. The same CT numbers were maintained after applying the
present invention compared with the original CT numbers. Noise
reduction using the present invention can be observed by the
smaller error bars compared with those in original images. The
percentage of noise reduction is shown in Table I.
TABLE-US-00001 TABLE I Percent noise reduction kVp Ca Water Tissue
Average 80 50% 51% 51% 51% 100 43% 37% 44% 41% 120 49% 53% 37% 46%
140 45% 45% 44% 45%
[0057] Overall, an approximately 50 percent noise reduction is
achieved using a conglomerate image and HYPR-based reconstruction.
Based upon the relationship between noise and radiation dose
(Dose-1/noise 2), this is equivalent to a factor of 4 dose
reduction given the same image noise. These results could be
interpreted as noise reduction given the same radiation dose, dose
reduction given the same image noise, or the combination of these
two.
[0058] CT numbers were well preserved using the present invention
and image noise was comparable to that of the composite image using
all photons, significantly reduced compared with standard filter
backprojection (FBP) algorithms. This breaks the trade-off between
image noise and energy bin size and/or numbers, which allows
flexibility to use optimal energy bins for best spectral
imaging.
[0059] Radiation dose reduction can be achieved with the reduced
image noise using the present invention. As addressed above, other
processing techniques for forming the conglomerate data set and
reconstructing the energy series of images are contemplated,
including Prior Image Constrained Compressed Sensing (PICCS) image
reconstruction. In this case, the conglomerate image may serve as
the so-called "prior image" of PICCS in PICCS-based
reconstruction.
[0060] As a simple special case, dual-energy CT, which is now
clinically used for stone composition differentiation, bone removal
in CT angiography, gout detection, and iodine quantitifcation, can
also benefit from this approach. In current clinical dual-energy CT
implementations, x-ray photons are approximately equally
distributed into the low and high energy scans. Each image set then
uses only one half of the total radiation dose delivered to the
patient and has higher noise compared with a conventional (single
energy) CT that uses the full dose to the patient. The method of
the present invention improves the noise property of each image set
to that of the conventional (single energy) full dose images, while
preserving energy-selective information for dual energy
processing.
[0061] Thus, dual-energy CT, which is one specific implementation
of the more general spectral CT, has been shown to provide
clinically useful diagnostic information beyond what is available
with conventional single energy CT. Using the present invention,
noise can be reduced in both the low- and high-energy images of a
dual energy CT exam. For example, HYPR-LR processing may be
conducted in low-and high-energy images before material
decomposition. Noise can also reduced after material-specific
processing is performed to subtract iodine signal (i.e. to create a
virtual, non-contrast-enhanced dataset). Because noise is reduced
in both the low-and high-energy images, any other dual energy
processing algorithms, such as those performed to simulate
monoenergetic images, also benefit from this invention.
Furthermore, it is contemplated that the above-described methods
may be used to process image data after material decomposition has
been performed. In such case, a "enhanced" material-specific image
may be generated in accordance with the above-described
processes.
[0062] In CT imaging, image noise is inversely proportional to the
number of available photons. Therefore, image noise in spectral CT
for each individual energy bin is determined by the number of
photons falling inside the energy range. The tradeoff between using
more energy bins (providing more energy specific measurements) and
decreasing image noise (providing more photons) has to be
considered. Noise is primarily determined by the conglomerate
image, which may use all the applied photons instead of only the
photons falling within a single energy bin. This has been
demonstrated using 2, 4, and 6 energy bins with a dual energy CT
patient exam (2 bins), conventional CT phantom scans (4 bins),
photon-counting p-CT (6 bins) and numerical simulations (6 bins).
More (greater than 2) energy bins could be used without
substantially increasing image noise. However, the number of energy
bins cannot be increased arbitrarily for a number of reasons.
First, the maximal number of energy bins is mainly determined by
the detector hardware. Second, based on Eqns. (2) and (3),
increasing the number of bins will slightly increase image noise,
especially when the number of bins becomes comparable to the number
of pixels in the convolution kernel. Third, the incremental benefit
of increasing the number of energy bins might not be significant
once it reaches a certain level. Currently, many photon counting
detectors operate using 2-8 energy bins. In this range, the
increase in image noise using more energy bins is negligible after
processing, and the tradeoff between more energy bins (more energy
specific information) and more photons (less noise) is practically
avoided.
[0063] In conclusion, it has been demonstrated the ability to
substantially reduce noise in spectral CT images by exploiting
information redundancies in the energy domain. The ability to
reduce image noise to the level of a conventional CT image that
uses all available photons eliminates the need to trade off image
noise (and hence, patient dose) and the number and width of energy
bins. This approach provides maximal flexibility for energy beam
optimization without paying a price in terms of increased noise or
dose.
[0064] The present invention has been described in accordance with
the embodiments shown, and one of ordinary skill in the art will
readily recognize that there could be variations to the
embodiments, and any variations would be within the spirit and
scope of the present invention. Accordingly, many modifications may
be made by one of ordinary skill in the art without departing from
the spirit and scope of the appended claims.
* * * * *