U.S. patent application number 12/150539 was filed with the patent office on 2012-04-19 for x-ray imaging with x-ray markers that provide adjunct information but preserve image quality.
Invention is credited to Kenneth F. Defreitas, Zhenxue Jing, Baorui Ren, Andrew P. Smith, Jay Stein.
Application Number | 20120093285 12/150539 |
Document ID | / |
Family ID | 41215019 |
Filed Date | 2012-04-19 |
United States Patent
Application |
20120093285 |
Kind Code |
A9 |
Ren; Baorui ; et
al. |
April 19, 2012 |
X-ray imaging with X-ray markers that provide adjunct information
but preserve image quality
Abstract
A method and an apparatus for estimating a geometric thickness
of a breast in mammography/tomosynthesis or in other x-ray
procedures, by imaging markers that are in the path of x-rays
passing through the imaged object. The markings can be selected to
be visible or to be invisible when the composite markings/breast
image is viewed in clinical settings. If desired, the contribution
of the markers to the image can be removed through further
processing. The resulting information can be used determining the
geometric thickness of the body being x-rayed and thus setting
imaging parameters that are thickness-related, and for other
purposes. The method and apparatus also have application in other
types of x-ray imaging.
Inventors: |
Ren; Baorui; (Andover,
MA) ; Smith; Andrew P.; (Lexington, MA) ;
Jing; Zhenxue; (Bedford, MA) ; Stein; Jay;
(Boston, MA) ; Defreitas; Kenneth F.; (Patterson,
NY) |
Prior
Publication: |
|
Document Identifier |
Publication Date |
|
US 20090268865 A1 |
October 29, 2009 |
|
|
Family ID: |
41215019 |
Appl. No.: |
12/150539 |
Filed: |
April 28, 2008 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
10723486 |
Nov 26, 2003 |
7831296 |
|
|
12150539 |
|
|
|
|
11271050 |
Nov 11, 2005 |
7577282 |
|
|
10723486 |
|
|
|
|
11827909 |
Jul 13, 2007 |
7616801 |
|
|
11271050 |
|
|
|
|
Current U.S.
Class: |
378/37 |
Current CPC
Class: |
A61B 6/583 20130101;
A61B 5/1075 20130101; A61B 6/544 20130101; A61B 6/502 20130101;
A61B 6/0414 20130101 |
Class at
Publication: |
378/37 |
International
Class: |
A61B 6/04 20060101
A61B006/04 |
Claims
1. A breast x-ray imaging method comprising: x-ray imaging a first
pattern of markers on or in a compression paddle and a patient's
breast immobilized between the paddle and an x-ray imaging receptor
to form a composite x-ray image in which both markers of the first
pattern of markers and the breast are imaged; computer-processing
the x-ray image to derive geometric information regarding markers
of the first pattern of markers imaged in said composite x-ray
image; computer-processing the geometric information to derive
thickness information related to a thickness of the immobilized
breast; and using the thickness information for at least one of
assessment of the imaged breast and further x-ray imaging of the
breast.
2. A method as in claim 1 in which the imaged markers of the first
pattern of markers are invisible when viewing the composite x-ray
image in clinical settings.
3. A method as in claim 1 in which at least some of the imaged
markers are visible in the composite image when viewed in clinical
settings.
4. A method as in claim 1 further including computer-processing the
composite x-ray image to remove thereform contributions thereto of
the imaged markers of the first pattern of markers.
5. A method as in claim 4 further including computer-processing the
composite x-ray image to remove therefrom essentially all
contributions thereto of the imaged markers of the first pattern of
markers.
6. A method as in claim 1 in which said first pattern of markers
comprises two or more first patterns at different locations
relative to said compression paddle, and said computer-processing
further comprising deriving geometric information regarding imaged
markers of each of said first patterns of markers and deriving
thickness information related to a thickness of the breast at each
of two or more different breast locations.
7. A method as in claim 6 including processing the information
related to thickness of the breast at each of two or more different
breast locations to derive information indicative of at least one
of tilt and deformation of the compression paddle.
8. A method as in claim 6 including storing said information
indicative of at least one of tilt and deformation in association
with stored image information describing at least the imaged
breast.
9. A method as in claim 1 in which said computer-processing to
derive geometric information regarding imaged markers of the first
pattern of markers comprises applying a correlation process to said
composite x-ray image.
10. A method as in claim 9 including selecting a region of interest
(ROI) in the x-ray image that is likely to contain at least a
substantial portion of the imaged first pattern of markers and
applying said correlation process only to the selected ROI.
11. A method as in claim 9 in which said correlation process
comprises a shift-and-add process.
12. A method as in claim 1 including further computer-processing
the image information to calculate density information
characterizing the immobilized breast.
13. A method as in claim 12 including further computer-processing
said density information to derive predictor information related to
cancer risk.
14. A method as in claim 13 including recording said cancer
predictor information in association with image information
regarding the imaged breast.
15. A method as in claim 12 including recording said density
information in association with image information regarding the
imaged breast.
16. A method as in claim 12 including displaying said density
information in association with a display of image information
regarding the imaged breast.
17. A method as in claim 1 in which said first pattern of markers
comprises spatially correlated markers.
18. A method as in claim 1 in which said first pattern of markers
comprises markers that are sufficiently transparent to the x-rays
used in said x-ray imaging to form imaged markers that are in the
range of general noise in said image.
19. A method as in claim 1 in which said x-ray imaging comprises
acquiring a composite x-ray image formed of a number of projection
composite x-ray images taken at different angles of an imaging
x-ray beam relative to the breast, and said computer processing
comprises deriving geometric information regarding markers of the
first pattern of markers imaged in said projection x-ray images and
computer-processing the geometric information to derive thickness
information related to thickness of the immobilized breast, and
processing information regarding the imaged breast from said
projection images to derive information describing tomosynthesis
slice image of the breast.
20. A method as in claim 1 including imaging a second pattern of
markers concurrently with said imaging of the first pattern of
markers such that said composite x-ray image further includes an
image of markers of the second pattern of markings, said second
pattern being at a second distance from the x-ray source that is
different from the distance of said first pattern from the source,
and wherein said computer-processing of the image information
further comprises deriving second geometric information regarding
the imaged markings of the second pattern of markings and deriving
therefrom distance information related to said second distance.
21. A method as in claim 20 in which said second pattern of
markings comprises two or more second patterns of markings that are
at different locations.
22. A mammography/tomosynthesis apparatus comprising: a breast
immobilizer comprising a breast platform and a compression paddle
with a first pattern of markers, said paddle and platform being
configured to move toward each other to position a patient's breast
between the paddle and platform and flatten and immobilize the
breast; an x-ray source on one side of the immobilizer and an x-ray
imaging receptor on the other side; a control selectively
energizing the x-ray source to emit an x-ray beam that traverses
the paddle, the immobilized breast and the breast platform and
impinges on the x-ray receptor, said receptor being configured to
provide in response image information describing a composite x-ray
image containing information regarding markers of the first pattern
of markers imaged in said composite x-ray image and regarding the
immobilized breast imaged in said image; a processor coupled with
the x-ray receptor to receive from the receptor the image
information describing the composite x-ray image, said processor
being configured to process the received image information to
derive geometric information regarding markers of the first pattern
of markers imaged therein and to process the geometric information
to derive thickness information related to thickness of the
immobilized breast; said processor further being configured to use
the thickness information in at least one of (i) assessment of the
breast imaged in the x-ray image and (ii) providing information
related to parameters involved in controlling further x-ray imaging
of the breast; and a display selectively showing a display related
to said x-ray image information.
23. An apparatus as in claim 22 in which said processor is
configured to further process said composite image information to
thereby remove therefrom essentially all contributions of the
imaged markers of the first pattern of markers and leave breast
image information.
24. An apparatus as in claim 22 in which said processor is
configured to further process said composite image information to
thereby remove therefrom contributions of the imaged markers of the
first pattern of markers.
25. An apparatus as in claim 25 in which said first pattern of
markers comprises two or more first sub-patterns of markers that
are at different locations relative to said compression paddle.
26. An apparatus as in claim 25 in which said processor is further
configured to derive geometric information regarding markers of
each of said first sub-patterns of markers imaged in said x-ray
image and to derive thickness information related to a thickness of
the breast at 32 each of two or more different breast
locations.
27. An apparatus as in claim 26 in which said processor is further
configured to process the information related to said thickness of
the breast at each of two or more different breast locations to
derive information indicative of tilt and/or deformation of the
compression paddle.
28. An apparatus as in claim 27 in which said processor is further
configured to store the information indicative or tilt and/or
deformation together with stored breast imaging information.
29. An apparatus as in claim 24 in which said processor is further
configured to select a region of interest (ROI) in the x-ray image
that is likely to contain at least a substantial portion of the
imaged markers of said first pattern of markers and to process only
the image information related to the ROI in deriving geometric
information regarding the imaged markers of the first pattern of
markers.
30. An apparatus as in claim 24 in which said processor is further
configured to process the breast image information to calculate
density information characterizing the immobilized breast.
31. An apparatus as in claim 30 in which said processor is further
configured to process said density information to derive predictor
information related to cancer risk.
32. An apparatus as in claim 31 in which said processor is further
configured to record said cancer predictor information in
association with said breast image information.
33. An apparatus as in claim 30 in which said processor is further
configured to record said density information in association with
said breast image information.
34. An apparatus as in claim 30 in which said processor is further
configured to cause said display to display said density
information in association with a display of said breast image
information.
35. An apparatus as in claim 22 in which said imaged markers of the
first set of markers are invisible when the composite x-ray image
is viewed in clinical settings.
36. An apparatus as in claim 22 in which the imaged markers of the
first set of markers are visible when the composite x-ray image is
viewed in clinical settings.
37. An apparatus as in claim 22 in which said first pattern of
markers comprises two or more first sub-patterns of markers that
are at different locations relative to said compression paddle.
38. An apparatus as in claim 37 in which said processor is further
configured to derive geometric information regarding markers of
each first sub-pattern of markers imaged in said x-ray image and to
derive thickness information related to a thickness of the breast
at each of two or more different breast locations.
39. An apparatus as in claim 22 in which said processor is further
configured to select a region of interest (ROI) in the x-ray image
that is likely to contain at least a substantial portion of the
imaged markers of said first pattern of markers and to process only
the image information related to the ROI in deriving geometric
information regarding the imaged markers of the first pattern of
markers.
40. An apparatus as in claim 22 in which said processor is further
configured to process at least the breast image information in said
composite x-ray image to calculate density information
characterizing the immobilized breast.
41. An apparatus as in claim 40 in which said processor is further
configured to process said density information to derive predictor
information related to cancer risk.
42. An apparatus as in claim 41 in which said processor is further
configured to record said cancer predictor information in
association with at least breast image information contained in
said composite x-ray image.
43. An apparatus as in claim 40 in which said processor is further
configured to cause said display to display said density
information in association with a display of breast image
information contained in said composite x-ray image.
44. An apparatus as in claim 22 in which said processor is
configured to apply a correlation process to said image information
describing the composite x-ray image to derive said geometric
information regarding imaged markers of said first pattern of
markers.
45. An apparatus as in claim 22 including a component with a second
pattern of markers situated between the x-ray source and an imaging
plane of said x-ray receptor and wherein said composite x-ray image
further contains information regarding markers of said second
pattern of markers imaged in said image, and wherein said processor
is further configured to process said information regarding imaged
markers of said second pattern of markers to derive information
regarding a position in space of said second pattern of
markings.
46. An apparatus as in claim 45 in which said component is a
magnification table between the breast and the breast platform,
said magnification table having an upper surface with said second
pattern of markings.
47. An apparatus as in claim 22 in which said apparatus comprises a
tomosynthesis apparatus, said composite x-ray image comprises a
plurality of tomosynthesis projection images at least one of which
contains information regarding markers of the first pattern of
markers imaged therein, said processor is configured to derive said
geometric information from the at least one projection image that
contains information regarding the markers and derive thickness
information therefrom, and said processor is further configured to
process the projection images to derive reconstructed slice images
representing slices of the breast having selected orientations and
thicknesses.
48. A method comprising: x-ray imaging a first pattern of markers
situated at a selected location relative to an object that is
between an x-ray source and an imaging plane of an imaging x-ray
receptor to thereby cause the receptor to form a composite x-ray
image in which both markers of the first pattern of markers and the
object are imaged; computer-processing the x-ray image to derive
geometric information regarding markers of the first pattern of
markers imaged in said x-ray image; and using the geometric
information to improve at least one of (a) further imaging of the
object and (b) assessment of an image of the object.
49. A method as in claim 48 in which the using step comprises
computer-processing the geometric information to derive thickness
information related to a position in space of said first set of
markings.
50. A method as in claim 49 in which the using step further
comprises using the geometric information for co-registration of
the x-ray image with other x-ray images.
51. A method as in claim 49 further comprising computer-processing
the image information and said geometric information to remove
contributions of markers of the first pattern of markers to the
composite image.
52. A method as in claim 48 in which said markers imaged in said
x-ray image are invisible when viewing the image in clinical
settings.
53. A method as in claim 48 in which said markers imaged in said
x-ray image are visible when viewing the image in clinical
settings
54. An x-ray apparatus comprising: an x-ray source, an x-ray
receptor, and an object support between the source and the
receptor; a control selectively energizing the source to emit an
x-ray beam that passes through an object on said support and causes
the receptor to provide an x-ray image of the object; a first
pattern of markers situated at a selected location in the path of
said beam thereby causing said image to contain contributions of
said markers; a processor configured to process the x-ray image to
derive geometric information regarding said contributions of said
markers, said processor further being configured to use the
geometric information to improve at least one of (a) further
imaging of the object and (b) assessment of an image of the
object.
55. An x-ray apparatus as in claim 54 in which the processor is
further configured to process the geometric information to derive
spatial information related to the location in space of said first
set of markings.
56. An x-ray apparatus as in claim 45 in which said processor is
further configured to process information describing at least some
of said x-ray image and to process said geometric information to
remove contributions of markers of the first pattern of markers to
the x-ray image.
57. A computer program product stored in a tangible computer
storage medium and comprising: program code to cause an x-ray
imaging system to image a first pattern of markers situated at a
selected location relative to an object that is between an x-ray
source and an imaging plane of an imaging x-ray receptor to thereby
cause the receptor to form a composite x-ray image in which both
markers of the first pattern of markers and the object are imaged;
program code to cause computer-processing of the x-ray image to
derive geometric information regarding markers of the first pattern
of markers imaged in said composite x-ray image; and computer code
to cause utilization of said geometric information.
58. A computer program product as in claim 57 further including
code to cause removal of contributions of said first pattern to
said composite image.
59. A computer program product as in claim 57 further including
code to cause removal of essentially all contributions of said
first pattern of markings from the composite image.
60. A computer code product as in claim 57 in which the markings
imaged in the composite x-ray image are invisible when viewing the
image in clinical settings.
61. A method of x-ray imaging a patient's breast comprising:
immobilizing a patient's breast between a compression paddle and a
breast platform that are in turn between an x-ray source and an
x-ray image receptor; determining a height of the compression
paddle using a first encoder and providing digital information
indicative of the determined height; determining tilt of the
compression paddle using a second encoder and providing digital
information indicative of the determined tilt; and using said
digital information indicative of compression paddle height and
tilt to improve at least one of x-ray imaging the patient's breast
and assessing images of the patient's breast.
Description
TECHNICAL FIELD
[0001] This patent specification relates to x-ray imaging and more
specifically to accurately finding the geometric thickness of an
object being x-rayed through including markers in the x-ray path
and computer-processing the composite markers/object x-ray image to
identify the markers and to use the resulting information to
calculate geometric object thickness and improve imaging.
BACKGROUND
[0002] Breast cancer and other breast lesions continue to be a
significant threat to women's health. X-ray mammography currently
is the most widely used tool for early detection and diagnosis, and
is the modality approved by the U.S. Food and Drug Administration
to screen for breast cancer in women who do not show symptoms of
breast disease. A typical x-ray mammography system compresses and
immobilizes a patient's breast on a breast platform positioned
between an x-ray source and an x-ray imager, and takes a projection
x-ray image (called here a conventional mammogram or simply
mammogram) using a collimated cone or pyramid beam of x-rays at
appropriate factors such as mA (current), kVp (voltage) or keV
(energy), and msec (exposure time). In the United States, typically
two views are taken of each breast, one from above (cranial-caudal,
or CC, with the image plane generally at a 0.degree. angle to the
horizontal and one from the side (mediolateral-oblique, or MLO,
with the image plane at an angle of typically around 45.degree. to
the horizontal). Different typical views may be taken in other
countries. The x-ray source typically is an x-ray tube operating at
or in the neighborhood of 25-30 kVp, using a molybdenum, rhodium,
or tungsten rotating anode with a focal spot of about 0.3 to 0.4 mm
and, in some cases, 0.1 mm or less. An anti-scatter grid between
the breast and the imager can be used to reduce the effects of
x-ray scatter. The breast is compressed to reduce patient motion
and also for reasons such as reducing scatter, separating
overlapping structures in the breast, reducing the x-ray thickness
of the imaged breast and making it more uniform, and providing more
uniform x-ray exposure. Traditionally, the imager has been a
film/screen unit in which the x-rays impinging on the screen
generate light that exposes the film. In the last several years,
mammography systems using electronic digital flat panel array
receptors have made significant inroads. A Selenia.TM. digital
mammography system with such a digital flat panel x-ray receptor or
imager is offered by Lorad, a division of the assignee hereof,
Hologic, Inc. of Bedford, Mass. See brochure "Lorad Selenia.TM."
Document B-BI-SEO US/Intl (5/06) copyright Hologic 2006. Digital
mammography has significant advantages and in time may fully
supplant film/screen systems. Additional information regarding
digital mammography systems and processes offered by the common
assignee can be found at <www.hologic.com>. Digital
tomosynthesis also has made advances and the assignee hereof has
exhibited breast tomosynthesis systems at trade shows and has
carried out clinical testing. It is a three-dimensional process in
which several two-dimensional projection views are acquired at
respective different angles but lower x-ray dose than conventional
mammograms, and are reconstructed into tomosynthesis slice views
that can be along any desired plane in the breast. For
tomosynthesis, the breast is still immobilized but may be
compressed to the same or lesser extent than in conventional
mammography. See, e.g., International Application WO 2006/058160 A2
published under the Patent Cooperation Treaty on Jun. 1, 2006 and
Patent Application Publication No. 2001/0038681 A1, PCT application
International Publication No. WO 03/020114 A2 published Mar. 13,
2003, U.S. Pat. Nos. 7,142,633, 6,885,724, 6,647,092, 6,289,235,
5,051,904, 5,359,637, and 4,496,557, and published patent
applications US 2004/0109529 A1, US 2004/0066884 A1, US
2005/0105679 A1, US 20050129172A1, and Digital Clinical Reports,
Tomosynthesis, GE Brochure 98-5493, 11/98. Reference markers can be
used in x-ray imaging for purposes such as checking the rotation
angle and unwanted shift of center of rotation of an x-ray source
and receptor (imager), and fiducial phantoms can be used in 3D
angiography to calibrate for irregular scan geometries. See, e.g.,
U.S. Pat. Nos. 5,051,904, 5,359,637, and 6,289,235, N. Navab, et
al., Dynamic geometrical calibration for 3D cerebral angiography,
SPIE Vol. 2708, pp. 361 370, and said PCT published application WO
03/020114 A2. A tomosynthesis system specifically for imaging
patients' breast is disclosed in commonly owned U.S. Pat. Nos.
7,123,684 and 7,245,694. The same system can be selectively used
for mammography and tomography, in the same or different
compressions of the patient's breast.
[0003] It is desirable to know the geometric thickness of the
immobilized breast in both film/screen and digital flat panel x-ray
mammography as well as in tomosynthesis in order to make
appropriate setting for the imaging procedure, such as settings for
the x-ray tube that control the x-ray beam. Knowing the breast
thickness can also help in quantitative assessments regarding x-ray
images of the breast, such as in assessing the nature and clinical
significance of x-ray attenuation properties of the breast. It can
also be used in tomosynthesis reconstructions such as to determine
the required reconstructed field of view or desired display field
of view. It has been proposed to use encoders to measure the
geometric height of a breast compression paddle and use the result
to estimate the geometric thickness of the breast but this process
typically has a relatively high error because of factors such as
tilting and geometric distortions of the paddle as it compresses
the breast, and because the encoders must be calibrated. The
information on the paddle height is oftentimes stored in the DICOM
header associated with the medical image, and this information is
used for several purposes, including the calculation of body part
radiation exposure. It also has been proposed to derive breast
thickness information from measuring the paddle compression force
versus time and to use the results to guide control factors such as
kV, mAs, and filter selection, as in commonly assigned U.S. Pat.
No. 7,123,684. The geometric thickness of any other body part being
x-rayed also can be of interest, and one process for basing an
estimate on non-contact ranging using ultrasound is discussed in
U.S. Pat. No. 4,597,094. It has also been proposed to use a
calibration phantom with embedded pellets of high-density material
serving as markers and to use the imaged markers in calibrating the
system, such as once a week. See U.S. Pat. No. 7,142,633 cited
above.
[0004] Proposals have been made for automated methods of estimating
breast density from mammograms, including volumetric and areal
estimates. One example, using digitized film mammograms, is given
in Br J Radiol. May 2006;79 (941):378-82 16632617. These methods
are understood to derive the volume of fibroglandular tissue in the
breast tissue for a cross-sectional area in a mammogram. The
accuracy of these methods is believed to be related to the accuracy
of measurement of the actual compressed breast. In addition,
information has been published proposing that breast density is
related to the breast cancer risk, See Cancer Epidemiology
Biomarkers & Prevention, Vol 13, 715-722, May 2004, American
Association for Cancer Research.
[0005] The patents and other publications identified above,
including the brochure "Lorad Selenia.TM." and said published
application WO 2006/058160 (corresponding U.S. patent application
Ser. No. 11/791,601), are hereby incorporated by reference in this
patent specification.
SUMMARY OF DISCLOSURE
[0006] In one non-limiting example, a breast imaging method
comprises x-ray imaging a pattern of markers on a compression
paddle and a patient's breast immobilized between the paddle and an
x-ray imaging receptor to form a composite x-ray image in which the
x-ray image of markers is composited with the x-ray image of the
breast, computer-processing the composite x-ray image to derive
geometric information regarding the imaged markers,
computer-processing the geometric information to derive thickness
information related to a thickness of the immobilized breast or
other information related to the position of the pattern of markers
in space or relative to system components, and using the thickness
or spatial position information for improving breast imaging. The
imaged markers may be invisible within the breast outline when
viewing the composite image in typical clinical settings or, in
alternative embodiments, they may be visible in the composite
image. The method can further include removing some or essentially
all the contribution of the markers from the composite x-ray
image.
[0007] As another non-limiting example, an apparatus comprises a
compression paddle with a pattern of markers that immobilizes a
patient's breast against a breast platform, an x-ray source on one
side of the paddle and platform and an x-ray imaging receptor on
the other side, a processor receiving image information from the
receptor related to a composite x-ray image of the markers and
breast in which the markers are not visible in a clinical setting,
said processor computer-processing the composite image to derive
geometric information regarding the markers therein and
computer-processing the geometric information to derive thickness
information related to thickness of the immobilized breast, and
said processor using the thickness information to improve imaging
the breast. The processor can further process the composite image
to completely or essentially remove some or essentially all
contribution of the imaged markers.
[0008] The method and the apparatus can be applied to both
mammography and tomosynthesis systems, and the term
"mammography/tomosynthesis" is used in this patent specification to
mean any one of a mammography system, a tomosynthesis system, and a
fusion system that can selectively carry one or both of mammography
and tomosynthesis, including while a patient's breast remains
immobilized.
[0009] In addition to or instead of being on the compression
paddle, similar markers can be placed on other objects. For
example, markers can be placed on the magnification table that is
placed between the normal breast platform and the breast to
increase the distance between the breast and the imaging plane and
thus magnify the breast image. Markers can be placed on the normal
breast platform if it desirable to calculate or confirm the spacing
between the breast platform and the image plane. In general,
markers can be placed on any object that is between the x-ray tube
and the image plane of the x-ray receptor to help calculate the
distance between the object and the image plane or other
objects.
[0010] A pattern of parallel lines, like the mm marks of a ruler,
can be used as the pattern of markings. However, many other
patterns of markings can be used and may be preferable depending on
the goals of the measurement. For example, a pattern can be dots or
a distributed collection of short increments of lines that will be
imaged as very small areas that are well dispersed in the x-ray
image. The distribution can even be pseudo-random, to further help
in making the imaged pattern invisible in clinical settings, so
long as the pattern has properties that can help find it when
imaged in the x-ray image; for example the pattern is known, or has
enough sharp edges that can differentiate the pattern elements from
background breast images through suitable edge detection of other
computer processing. If the markers in a pattern are elongated,
they can be at any desired orientation.
[0011] While this patent specification discusses, as one
alternative, patterns of markings that when imaged are invisible in
the composite x-ray image when viewed in typical clinical settings,
it also explains an alternative in which the imaged markers are
visible in the composite image when viewed in clinical settings. In
he first alternative, the x-ray attenuation properties of the
patterns of markers can have values that are insufficient to make
the imaged markers visible in clinical settings but sufficient to
allow them to be identified through computer processing. In the
second alternative, the pattern of markers may have sufficient
x-ray attenuation to make the imaged markers visible in the
composite image in typical clinical settings; however image
processing can process the composite image to remove some or
essentially all contribution of the imaged markers. Even when the
imaged markers are invisible in the composite image in typical
clinical settings, their contributions to the composite image can
be removed through further computer processing to reduce possible
interference with image processing such as CAD (computer aided
detection). Another example of the patterns of markers arises in
systems capable of tomosynthesis imaging. Tomosynthesis
examinations acquire of raw, or projection images, and subsequently
reconstruction them into tomosymthesis slice images that can be
desired orientations and represent breast slices of desired
thicknesses. Because the radiological assessment is typically made
from the reconstructed slice images, patterns of markings and
analysis as described in this patent specification are useful even
if they are visible in the tomosynthesis projection images. This is
because their contrast and visibility will be greatly reduced in
the reconstructed slices that encompass the breast volume, because
the patterns are physically located outside the breast. Because
these examples relax the requirement for complete invisibility,
additional uses become practical. In a typical digital mammography
or tomosynthesis examination, the setting of x-ray exposure
techniques are determined through the use of Automatic Exposure
Control (AEC) methods, which typically use a low dose x-ray
pre-pulse. The low doses employed in the pre-pulse means that the
visibility of patterns of markings is reduced and so their contrast
will need to be increased accordingly. Patterns of markings and
methods according to this patent specification can be used when the
appearance of the imaged markers is reduced through image
processing, or when the imaged markers do not appear in the breast
images of interest because those images are tomosynthesis
reconstructed images representing slices of the breast while the
imaged markings would mainly influence reconstructed images (if
any) of slices outside the breast volume.
[0012] As yet another non-limiting example, computer software
stored in a tangible storage medium can control an x-ray system to
carry out implementation of the process described in this patent
specification. Information such as breast thickness, paddle height
and paddle deformation so obtained can be stored in the DICOM
header associated with the medical image for subsequent uses.
[0013] The process and apparatus are not limited to applications
for breast imaging but are applicable to other x-ray procedures as
well.
BRIEF DESCRIPTION OF THE DRAWINGS
[0014] The features of the subject matter of this patent
specification can be more readily understood from the following
detailed description with reference to the accompanying drawings
wherein:
[0015] FIG. 1 is a schematic sectional view illustrating a geometry
useful in estimating a thickness of a patient's breast immobilized
between a compression paddle that has a pattern of markers and a
breast platform, when imaged with x-rays in accordance with one
embodiment of the present patent specification;
[0016] FIG. 2 illustrates a two-dimensional projection, such as a
mammogram or a tomosynthesis projection image, containing a pattern
of markers in the form of parallel bars or lines, where the imaged
pattern has been emphasized for illustrative purposes but would be
invisible in a clinical mammogram or other projection x-ray
image;
[0017] FIG. 3 illustrates an effect of shift-and-add correlation
processing of the pattern of FIG. 2 where the shift distance does
not equal the pitch, or separation distance, between the markers in
the image;
[0018] FIG. 4 illustrates an effect of shift-and-add correlation
processing of the pattern of FIG. 2 where the shift distance equals
the pitch of the markers in the image;
[0019] FIG. 5a illustrates a ruler pattern that can be used to
verify an example of the process disclosed in this patent
specification; FIG. 5b illustrates a top view of a compression
paddle in current public use in mammography systems offered by the
assignee hereof, with automatic exposure control (AEC) markers that
are not visible in a clinical image; and FIG. 5c illustrates the
compression paddle with an added example of a pattern (a ruler
pattern as in FIG. 5a) for use with an embodiment of the disclosed
process;
[0020] FIGS. 6a and 6b pertain to x-ray images taken with a
compression paddle and ruler as in FIG. 5c, compressing a 4.5 cm
thick BR50/50 phantom (BR50/50 is material simulating a breast
comprised of 50% fatty and 50% glandular tissue, such as sold by
Cirs Inc Model 014A Mammography Phototimer Consistency Testing
Slabs, available from CIRS in Norfolk, Va., see
http://www.cirsinc.com/pdfs/014Acp.pdf) in the case of FIG. 6a but
a 4.5 cm thick cadaver breast phantom in the case of FIG. 6b. In
each of FIG. 6a and 6b: strip 1 illustrates a raw x-ray image of
pattern and phantom/breast; strip 2 illustrates a pattern-corrected
image obtained by removing the image of the pattern from the image
of strip 1; strip 3 illustrates a difference image obtained by
subtracting the strip 2 image from the strip 1 image; strip 4
illustrates a shift-and-add image of raw image 1 where the shift is
at the correct distance and thus the imaged pattern of bars appears
in the strip 4 image as an interference signal that is being
detected; and strip 5 illustrates a shift-and-add image of image 2
where the shift also is at the correct distance but no interference
signal is visible, demonstrating that the pattern that was present
in strip 1 has been removed through pattern correction
processing;
[0021] FIG. 7a is a graph illustrating an example of amplitude
variations of an interference signal as in FIG. 6b (strip 4) vs.
estimated compression paddle height above a breast plate at normal
1.times. clinical x-ray dose. The true height of the compression
paddle was 4.5 cm, and the measured height using the methods
disclosed in this patent was 4.49 cm, identified through the peak
of the curve in FIG. 7a. The interference signal amplitude peaks at
a point within 0.01 cm of the paddle height. FIG. 7b illustrates
the interference signal from a specific shift-and-add image in the
same example, as an averaged signal plot in the spatial domain in
the upper plot and in the frequency domain in the lower plot, after
frequency analysis of the spatial domain plot through FFT (Fast
Fourier Transform) analysis;
[0022] FIG. 8a is a graph illustrating an example of amplitude
variations of an interference signal as in FIG. 6b (strip 4) but
with a 10.0 cm thick cadaver phantom vs. compression paddle height
at 1/15.sup.th of a normal 1.times. clinical x-ray dose. The true
height of the compression paddle was 10.16 cm, and the measured
height using the methods disclosed in this patent was 10.13 cm. The
interference signal amplitude peaks at a point within 0.2 cm of
phantom thickness. FIG. 8b illustrates the interference signal from
a specific shift-and-add image in the same example, as an averaged
signal plot in the spatial domain in the upper plot and in the
frequency domain in the lower plot, after frequency analysis of the
spatial domain plot through FFT (Fast Fourier Transform)
analysis;
[0023] FIG. 9 is a partial side view of a mammography system in
which a patient's breast is immobilized between a compression
paddle and a breast platform and is imaged on a flat panel digital
imager, with or without the use of an anti-scatter grid, in
accordance with one embodiment of the system and method disclosed
in this patent specification; and
[0024] FIG. 10 is a block diagram illustrating main portions of a
system in which the disclosed process is used.
[0025] FIG. 11 illustrates an example using two sets of markers, at
different heights above an imaging plane (detection layer) that can
facilitate calculation of the heights, and illustrates an x-ray
image showing the imaged markers (that would not normally be
visible in a clinical image);
[0026] FIG. 12 is a flowchart illustrating steps in a process of
estimating height of objects with markers;
[0027] FIG. 13 illustrates shift-and-add according to the flowchart
of FIG. 12;
[0028] FIG. 14 illustrates two different patterns of markings: a
pattern "a" in the form of a regular pattern of parallel short
lines and a pattern "b" that can be arbitrary and can even be
random or pseudo-random, so long as it has characteristics that can
differentiate it from a background x-ray image to which it
contributes; and
[0029] FIG. 15 is a schematic illustration of another way of
measuring both height and tilt of a compression paddle.
DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS
[0030] This patent specification describes methods and systems in
which the geometric thickness of a body part such as the breast
that is being x-rayed is accurately determined and used to improve
imaging in a manner that does not inconvenience the patient or the
health professional. It also describes similar processes to
determine other thicknesses, heights, or distances between objects
in x-ray examination procedures.
[0031] In describing preferred embodiments illustrated in the
drawings, specific terminology is employed for the sake of clarity.
However, this patent specification is not intended to be limited to
the specific terminology so selected and it is to be understood
that each specific element includes all technical equivalents that
operate in a similar manner. In addition, a detailed description of
known functions and configurations will be omitted when it may
obscure the subject matter of the invention described in the
appended claims.
[0032] As illustrated in the example of FIG. 1, a mammography or
tomosynthesis system immobilizes a patient's breast 12 between a
compression paddle 10 and a breast platform 14. An x-ray source 16
when energized emits x-rays from a focal spot 18 that pass through
paddle 10, breast 12 and platform 14 and are imaged at an imaging
plane of an imaging receptor 20 that is below platform 14 and may
for some x-ray procedures be separated therefrom by an anti-scatter
grid not shown in FIG. 1. A pattern 22 of five markers "a, b, c, d,
e" is formed in or on paddle 10 and is imaged by the x-rays as
respective imaged markers "A, B, C, D, E" in the projection breast
image formed at the imaging plane of receptor 20. The overall
geometry of a mammography/tomosynthesis system to which the
arrangement of FIG. 1 can be added is described in said published
application WO 2006/058160.
[0033] It can be appreciated that if the geometry of markers 22 is
known and the geometry of imaged markers 24 is measured in the
x-ray image, trigonometric calculations can yield the distance
between markers 22 and the imaging plane. The distance between
markers 22 and the bottom of compression paddle 10, and the
distance between the imaging plane and the top of breast platform
are known. These distances are sufficient to calculate the
geometric thickness of immobilized breast 12. An additional
challenge arises in cases where it is not acceptable or desirable
to have an easily visible contribution of the imaged markers to the
composite x-ray image. For example, it may not be practical in at
least some cases to have imaged markers visible in the breast image
as they may detract from the clinical value of the x-ray image.
[0034] The new approach disclosed in this patent specification
involves, as one alternative, forming a composite x-ray image of
tissue and superimposed markers such that the markers are invisible
in the composite image in a clinical setting but can be identified
though computer-processing of the composite image. As another
alternative, the markers are visible in the composite image. In
either or both alternatives, the disclosed process and system
further provide a way to remove contributions of the imaged markers
from the composite image, if desired. "Invisible" in this patent
specification means that the presence of imaged markers are not
visible to an observer within the outline of a patient's breast in
a realistic anatomical breast background under clinical x-ray dose,
in typical clinical settings and in the type of visual examination
of breast x-ray images that is done in conventional clinical
screening and/or diagnosis. The approach disclosed in this patent
specification is applicable to conventional mammography using flat
panel or other digital imaging receptors, to digitized film, to
tomosynthesis imaging using such receptors, and to other types of
x-ray procedures and other x-ray imaging receptors, where it is
desirable or helpful to accurately estimate the geometric thickness
of the body or body part being x-rayed and to use the results to
improve imaging or for other purposes.
[0035] A non-limiting example of the new approach uses one or more
patterns of markers 22 that produce imaged markers 24 whose
contrast relative to the breast image, or image of other body
parties, can be so low that the imaged markers would not be seen in
conventional assessment of x-ray images but can be detected through
computer-processing of the images using techniques such as
correlation analysis. This is made possible by using markers 22
that are in patterns with appropriate properties such as spatial
distribution or other spatial and x-ray density properties as, for
example, repeated line bars at known pitch. Correlation analysis of
the imaged markers 24 then can allow detection of the position and
pitch of imaged markers 24 in the composite markers/breast image.
Once the geometry of the imaged markers 24 has been found, further
computer processing can remove the contributions of markers 24 from
the composite image. This removal can be so effective that an image
that should be flat is in fact flat despite the presence of markers
22 at paddle 10. "Flat image" here means an x-ray image taken with
a uniform phantom between paddle 10 and image receptor 20, i.e.,
and image that should have a uniform gray level within system
tolerances and when viewed in typical clinical setting appears to
be flat.
[0036] The effectiveness and robustness of the new approach can be
demonstrated by experiments with a common ruler made of transparent
plastic and having small painted black bars marking the millimeters
(mm). In these experiments, the ruler is placed on paddle 10, and
the ruler's mm marks serve as a pattern of markers 22. Breast
phantoms of geometric thickness from 0 cm to 10 cm are immobilized
between the paddle and the breast platform and are imaged under
clinical x-ray dose levels in a mammography unit such as the system
available in this country under the tradename Selenia.TM. from the
assignee hereof, Hologic, Inc. of Bedford, Mass. Such experiments
can demonstrate that geometric breast thickness can be calculated
with errors within a .+-.1 mm range. Under clinical x-ray dose and
realistic anatomical breast background, the imaged mm marks of the
ruler are invisible to an observer. In a flat field image with a
uniform phantom, the ruler image is faint but perceptible. However,
additional computer processing as disclosed in this patent
specification can remove this residue of the ruler image to thereby
leave no visible artifacts in both flat field images and breast
images. In similar tests with 1/10th to 1/20th of normal clinical
x-ray dose, the ruler marks can still be identified through
computer processing and breast thickness can be measured to within
.+-.2 mm error range. This can allow geometric breast thickness to
be computed from an x-ray image obtained with the AEC (automatic
exposure control) scout shot, the very low dose x-ray shot
typically taken for automatic exposure control purposes in
mammography units such as the aforementioned Selenia.TM. system
before actual beast imaging. The breast thickness calculated from
the AEC shot information can then be used for more accurate AEC
procedures to achieve better quality breast images than with less
accurately determined geometric breast thickness, and/or for other
purposes.
[0037] In the illustrative examples of FIGS. 1-4, five tiny bar
markers 22 are used at paddle 10, uniformly spaced at a known pitch
"d." They project in the image formed at receptor 20 as five
similarly shaped imaged markers 24 that also are uniformly spaced
but at an unknown pitch "D". As illustrated in FIG. 1, "h" is the
distance between the focal spot 18 in x-ray source 16 and
compression paddle 10, "H" is the distance from focal spot 18 and
the image plane of receptor 20, the breast thickness is designated
"breast_thickness," and "s" designates the distance between the
breast platform and the image plane. Parameters "d" and "H" and "s"
are known, and parameter "D" can be determined as explained in this
patent specification. Parameter "h" then can be calculated from the
relationship h=(d/D)H, and breast thickness can be calculated as
(breast_thickness=H-h-s). In this example, the centerline of the
x-ray beam is normal to an centered at the image plane 20; however,
if the centerline is at a different angle and point of incidence in
the imaging plane, as in the case of some tomosynthesis projection
images, a uniform ruler pattern as in FIG. 1 may be geometrically
distorted. However, the distortion can be calculated at least
approximately from knowing system geometry at the time the
projection image is taken, e.g. from various encoders for the
position of the x-ray source and from the construction of the
system, so these geometric distortion can be effectively accounted
for in the calculations disclosed in the patent specification.
[0038] In this alternative example imaged markers 24 would not be
visible in a realistic breast image in a clinical setting, but to
assist in the discussion below they are illustrated as visible in a
composite marker/breast image in FIG. 2. Correlation processing can
be carried out by shifting the image of FIG. 2 by a specified
distance, e.g. to the right, adding the original and shifted
images, then shifting the image of FIG. 2 again by the same
distance and adding each shifted image to the result of the
previous shift-and-add, etc., as long and the original and the last
shifted images partially overlap. The shifting in general is not by
an integer number of pixels so appropriate image interpolation can
be carried out to generate the final shift-and-added image. The
result of this shift-and-add process, or a normalized version
thereof, can be expressed as amplitude related, for example, to an
average attenuation or gray level of pixels in the summed images.
This shift-and-add process is repeated for each of a number of
different shift distances, e.g., for each of a progression of
distances that differ in small steps and are in a reasonable range
related to expected breast thickness ranges and taking into account
possible magnification imaging in which the breast is spaced from
platform 14 by a selected distance and, if desired, possible
geometric distortions of pattern 22 because of factors such as a
non-normal angle between the x-rays that form the image of a marker
and the imaging plane. If the shift distance matches the actual
pitch "D" of the imaged markers 24, the amplitude would be at a
peak because the imaged markers 24 in effect line up and add while
the rest of the composite x-ray image is more random and does not
increase in amplitude as much. If the shift distance does not match
the pitch "D," the amplitude resulting from addition is less
because markers 24 are not aligned or at least not as aligned.
[0039] The shift-and-add process has been described qualitatively
above, but in actual practice it can be performed by processing the
pixel values of the composite markers/breast image. Consider the
pixel values derived from receptor 20 representing pixels that are
all within the middle imaged marker 24. After a single
shift-and-add by "D," the pixel values contributed by the same
marker 24 will be doubled; after another shift-and-add they will be
tripled; and so on. However, if the shift is by a distance
different from "D," at least some of the pixel values for the
marker will not be doubled--they will be added to pixel values
representing only the breast image. The pixel values for the entire
composite image can be added or averaged, and the resulting
amplitude should be highest for shift-and-add processes where the
shift distance was "D." Preferably, the shift-and-add process is
applied to one or more regions of interest (ROI) likely to contain
imaged markers 24 rather than to the entire x-ray image. Many
mathematical techniques are known for use in searching for weak but
periodic signals, e.g. signals that are below a noise level of
their environment. Such techniques generally involve some form of
correlation analysis, and many correlation processes known in
mathematics may be used in place of the shift-and-add example
discussed in more detail in this patent specification. However, to
the inventors' knowledge no such techniques have been applied in a
process and apparatus as disclosed in this patent specification,
e.g., for accurate determination of the geometric thickness of
breast or other tissue being x-rayed and to use of the resulting
information according to this patent specification.
[0040] A single pattern of markers can be used, or several patterns
can be used at the same level or at different levels. While in
these experiments a ruler mm marks are used as an example of a
pattern, other patterns can be used so long an they have known
geometries and are correlated, e.g., repeating at known pitch or
otherwise correlated. While FIG. 1 illustrates a pattern 22 of only
five bars, typically many more bars or other marks would be used.
Only a selected part of the composite image can be processed that
is likely to contain a significant portion of the pattern or
patterns. For example, if one or more patterns are placed at the
central portion of paddle 10, say 1 cm from the paddle edge that is
pressed against the patient's chest wall, then only a strip of the
x-ray image that is likely to contain an image of that portion of
paddle 10 can be computer-processed as described above. In
addition, in cases where paddle tilting is deliberately provided,
patterns or sub-patterns of markers 22 can be provided at different
places of paddle 10, and the thickness of the immobilized breast
can be calculated for the portion of the breast under each pattern
and interpolated for other portions of the breast. The resulting
information can be used to set exposure parameters and possibly for
breast image processing purposes specific to respective portions of
the imaged breast.
[0041] FIG. 3 shows four shifts of the image of FIG. 2 where each
shift is by a constant distance that is not equal to the pitch "D,"
and FIG. 4 is a similar illustration but for shifts by a distance
"D." As seen, the imaged markers 24 are misaligned in FIG. 3 but
not in FIG. 4, where the second marker from the left is the sum of
marker 2 of the original image and marker 1 of the image after one
shift by "D," etc.
[0042] An illustrative but non-limiting way to implement the
process in practice involves the following basic steps: [0043] a.
Compression paddle 10 is painted with markers 22 in a line bar
pattern with a known bar pitch, e.g., 1 mm, as in a normal plastic
ruler. Each line bar is about 6 mm long and is extends in a
direction normal to the chest wall such that the bars are repeated
along the chest wall direction in a mammographic or tomosynthesis
procedure. Each paint line is thin, e.g., about 0.1 mm to 0.2 mm,
for better correlation analysis. The contrast of the paint lines
preferably should be low enough such that the imaged markers 24 in
the composite marker/breast image are invisible when the composite
x-ray image is viewed under typical clinical conditions; [0044] b.
Because the height of paddle 10 can be known to within about .+-.1
cm from conventional encoder information supplied in digital form
to the system computer in a mammography unit such as the
Selenia.TM. system, it can be calculated through basic
trigonometric relationships approximately where the imaged markers
24 will be in the composite marker/breast x-ray image and
approximately what the pitch will be of imaged markers 24. These
calculations identify a region of interest (ROI) in the composite
image that will likely contain imaged markers 24 or at least a good
portion of the markers, and also identify a likely shift distance
for the correlation analysis or at least a range of likely shift
distances. In addition, if angular encoding is used to measure tilt
of the paddle (see FIG. 15 and description thereof below) and
several sub-patterns are used (as in FIG. 5c), the tilt information
can be used to further refine the estimate of the likely location
in the image of imaged markers of each sub-pattern, using known
trigonometric relationships; [0045] c. In one example of
correlation image analysis, this ROI is shifted incrementally along
the chest wall direction that is normal to line bars, with
different shift amounts within the expected range, and the image
resulting from the additions is checked for the line bar
interference pattern; [0046] d. Once the shift amount is equal to
the line bar pitch, the resulting image will show strongly enhanced
line bar patterns, say a factor of 50-100 times stronger relative
to the background. In addition to shift-and-add along chest wall
direction, all pixels along the 6 mm long line bar as imaged in the
composite image are added. This can give another factor of 100-300
times more signal in the final line profile; [0047] e. In frequency
domain analysis, the strongest interference signal can be picked up
at the expected frequency, and the corresponding shift amount can
be found that is equal to the line bar spacing. From the measured
pitch "D" of the imaged ruler markers at unknown height and the
known pitch "d" of the ruler, the system magnification ratio can be
calculated, and then the paddle height or breast thickness is
derived at the ROI region. If multiple clusters or sub-patterns of
line bars are placed at paddle 10, the analysis can be repeated to
get the paddle height profile and breast thickness at each ROI
within the whole detector area, and to calculate paddle tilt angle
or regional deformations as well; [0048] f. The process
incorporates principles similar to those used in the design of a
"phase lock amplifier" in analog electronics to identify a weak
electronic signal of known frequency in the time domain. Similar
principles can be used in the spatial domain as a form of a "phase
lock amplifier" to measure a spatial frequency in an x-ray image;
[0049] g. Once the pitch of imaged markers 24 and the paddle height
are calculated, the location and the signal magnitude of imaged
markers 24 in the composite marker/breast image can also be found
through conventional trigonometric relationships. Additional
artifact correction algorithms can therefore be applied to remove
the residue signal of the ruler marks (imaged markers 24) from the
composite image, which will make the ruler marks become completely
or essentially invisible even in the flat field image of the ruler.
After such removal of imaged marker contributions, the x-ray images
can be more suitable for processing such as CAD (computer aided
detection) analysis.
[0050] FIG. 5a shows a magnified ruler pattern that can be used in
the experiments referred to above. As seen, the ruler pattern has
mm marks. The pattern was printed, using a laser printer, onto a
thin sheet of plastic, but can be provided by other printing
techniques, or can be engraved into the paddle, or can be formed in
some other way, so long as it has the required x-ray properties
that would cause imaging the pattern into a composite
pattern/breast image as an imaged pattern that is invisible within
the breast outline when viewing the composite image in clinical
settings or, in alternative embodiments, the imaged pattern is
visible if the composite image is viewed in clinical settings.
Optionally in either alternative, contributions of imaged markers
can be removed from the composite x-ray image through further
computer processing. FIG. 5b shows a typical compression paddle 10
that can be used with a Selenia.TM. mammography system, with
painted marks related to AEC features, which AEC marks are not
visible in the breast x-ray image in clinical settings though they
can be clearly seen on the paddle. FIG. 5c illustrates the paddle
of FIG. 5b with an attached plastic sheet containing ruler patterns
similar in principle to that of FIG. 5a but the pattern in FIG. 5c
is in the form of nine different sub-patterns of ruler marks, where
both the AEC marks and the imaged ruler marks of the sub-patterns
are visible on the paddle but can be invisible when imaged in a
breast x-ray image in a clinical setting. For clarity, the marks of
the nine sub-patterns shown in FIG. 5c are coarser and the parallel
lines are fewer in number in each sub-pattern than they typically
would be in actual practice. While the sub-patterns are shown as
extending only in the lateral direction, in practice the
sub-patterns may extend in different directions that can differ
between sub-patterns. And, the sub-patterns need not be in the form
of ruler marks, as discussed in more detail below and different
sub-patterns can have different marks in different
arrangements.
[0051] FIGS. 6a illustrates results of experiments using a paddle
10 with a superimposed ruler pattern as seen in FIG. 5c
immobilizing a 4.5 cm BR50/50 breast phantom, and FIG. 6b is
similar but illustrates results when using a 4.5 cm cadaver breast
phantom, in each case with the use of a Selenia.TM. mammography
system. Strip 1 of FIG. 6a illustrates an x-ray image (actually a
strip taken from an image) that contains the combined contribution
of the breast phantom and the ruler mm marks serving as a pattern
of markers 22. The imaged markers 24 are not seen in strip 1 in a
typical observation in clinical settings. Strip 2 illustrates the
same image but with imaged markers 24 removed using the process
disclosed in this patent specification. No difference can be seen
between strips 1 and 2 in typical clinical setting observations.
Strip 3 illustrates a difference image obtained by subtracting the
images of strips 1 and 2 from each other. As expected, the imaged
markers 24 can be seen in this difference image, demonstrating that
there is information about markers 24 in the image of strip 1.
Strip 4 illustrates the result of a shift-and-add operation on the
image that contains strip 1, using a shift distance that matches
the actual pitch of imaged markers 24. The shift-and-add operation
reinforces and emphasizes imaged markers 24, making them visible in
the composite x-ray image and thus in another way demonstrating
that they are present in the composite image that contains strip 1.
In practice, this process detects the interference signal due to
the presence of markers 24 in the composite image of strip 1. Strip
5 illustrates the result of a shift-and-add operation on the image
of strip 2, carried out in the same way as the one that produced
strip 4. However, imaged markers 24 are not seen in strip 5,
demonstrating that image 2 was in fact pattern-corrected by
removing essentially all the contributions of the imaged marker 24
from the composite image of strip 1. FIG. 6b has corresponding
respective strips 1-5 but pertains to imaging a cadaver breast
phantom.
[0052] FIG. 7a shows a plot of an amplitude of the interference
pattern of imaged markers 24 (seen in FIG. 6a, strip 3) versus
different shift distances and also shows the measured breast
thickness values, in the case of using a 4.5 cm thick cadaver
breast phantom and normal x-ray imaging dose for mammography. Each
of the data points is for a different shift distance. The peak
signal amplitude corresponds to calculated breast thickness. As
seen in FIG. 7a, the calculated breast phantom thickness is within
0.01 cm of the 4.5 cm phantom thickness. The upper plot in FIG. 7b
illustrates for the geometry of FIG. 7a an averaged interference
signal in the spatial domain of imaged markers 24 from one specific
shift-and-add image shown in strip 4 of FIG. 6b. The lower plot in
FIG. 7b shows results from FFT (Fast Fourier Transform) analysis,
including a peak signal at the expected spatial frequency. FIGS. 8a
and 8B are similar plots, respectively, but for the case of using a
10.0 cm thick cadaver phantom and 1/15.sup.th of normal x-ray dose
for mammography. The amplitude of the interference signal
(peak-to-background) is reduced about 40 times relative to the case
of FIG. 7a (from about 0.2 to about 0.005) but the process
disclosed in this patent specification can still detect this weak
pattern signal and calculate breast phantom thickness to within 0.2
cm of the 10.0 cm phantom thickness, demonstrating the high
robustness of the process.
[0053] FIG. 9 illustrates certain components of a system that can
be used for mammography or tomosynthesis, in each case using the
process of determining breast thickness and using the result to
improve breast imaging. An x-ray source 1 is at one end of a
generally C-shaped frame 7 and a flat panel digital x-ray imaging
receptor 5 is at the other end. X-ray source 1 includes a
collimator schematically illustrated at 40 to confine an x-ray beam
30 emitted from source 1 to a desired footprint at receptor 5,
typically no larger than the area of receptor 5 and preferably just
enough to image a patient's breast 3 or at least a selected part
thereof, as compressed toward receptor 5 by a compression paddle 2
mounted on an arm 6 that in turn mounts to frame 7. A lower
platform 11, often called a breast tray or platform, is immediately
below the breast, and a scatter-reducing grid 4 is between breast
tray 11 and x-ray receptor 5 and is housed in the same enclosure 12
with the receptor. As is known in the art, frame 7 can rotate
between horizontal and vertical directions of x-ray beam 30. In use
for a CC view, paddle 2 and its supporting arm 6 are moved up,
breast 3 is positioned on tray 11 and compressed by bringing paddle
2 down as needed. With suitable collimation by collimators 40
(which typically collimate in two directions, of which only one is
illustrated in FIG. 9), beam 30 from source 1 images the breast
onto receptor 5 and the resulting electronic image information is
transmitted to a viewing station 112 (FIG. 10). The image typically
is rectangular. Preferably, the collimation is such that beam 30
illuminates an area of receptor 5 just large enough to show the
image of breast 3, or at least a selected part thereof. Different
sizes and shapes of paddles 2 can be mounted to arm 6, and the
paddle can be selectively positioned off-center relative to
proximal edge 5a of receptor 5 (the left edge in FIG. 9). For
mammography, grid 4 can be in the position shown in FIG. 9. For
tomosynthesis data acquisition, grid 4 can be retracted to the
right of FIG. 9 and out the path of x-rays 30, and a number of
projection breast images can be taken at different angles of the
x-ray beam relative to the breast and processed to derive
tomosynthesis slice images. Paddle 2 of FIG. 9 can be provided with
one or more patterns or sub-patterns of markers 22 as with paddle
10 of FIG. 1. Breast platform 11 and receptor 5 of FIG. 9
correspond to platform 14 and receptor 20 of FIG. 1.
[0054] A system that embodies an example of the new approach to
finding and using geometric breast thickness is illustrated in
block diagram form in FIG. 10. The system can be a mammography
system, or a tomosynthesis system, or a fusion system that can be
used for either or both purposes. A beast immobilizing device 102
is between an x-ray source 104 and an image receptor/grid unit 106.
Device 102 can comprise paddle 10 and breast platform 14 of FIG. 1,
or paddle 2 and breast platform 11 of FIG. 9. Unit 104 can comprise
source 16 and focal spot 18 of FIG. 1, or source 1 and collimator
40 of FIG. 9. Unit 106 can comprise receptor 20 of FIG. 1 or the
combination of receptor 5 and grid 4 or FIG. 9. A control unit 108
is coupled with units 102, 104 and 106 to control their operation
and to send and receive information regarding their operation. An
image processor 110 is coupled with unit 106 to receive image data
therefrom and exchange control and other information therewith, and
also is coupled with unit 108 to exchange control and other
information. A display/operator console unit 112 is coupled with
image processor 110 and control 108 to exchange display and other
information therewith, and to display images provided by unit 112
and receive user input through appropriate interface devices. The
units illustrated in FIG. 10 can include the functionalities of the
corresponding units of the commercially available Selenia.TM.
mammography system as well as the additional processing
functionality to carry out the process of determining breast
thickness and using the result for AEC and/or other purposes. The
additional processing functionality can further include breast
density calculations that use information regarding the geometric
thickness of the breast as imaged at one or more regions of the
imaging receptor.
[0055] An initial application of the disclosed method and apparatus
is to provide accurate estimates of breast thickness in mammography
and/or tomosynthesis systems using the AEC scout view before the
main exposure and to use the thickness information to set x-ray
technique, to provide accurate estimates of breast thickness for
more accurate breast density calculations, and for other purposes
such as control over tomosynthesis slice image reconstruction and
display. In tomosynthesis, the thickness information from a
projection image can be used in AEC for other projection images
and/or in AEC for a subsequent mammogram taken in the same breast
compression in the same system. Similarly, the thickness
information from a mammogram can be used in AEC for taking
subsequent tomosynthesis views or other x-ray images taken in the
same breast compression in the same system. Another use of the
process and apparatus for compositing invisible or visible markers
into an x-ray image and if desired removing such imaged markers
through further processing is in digital encoding of selected
information into an x-ray image. The composited marker information
can be used for a wide variety of purposes such as identifying the
patient, the procedure for acquiring the data for the image,
encoding other information regarding the image, or for providing
any other type of information embedded by way of invisible markers.
A similar process and apparatus can be used in other x-ray
procedures to determine the geometric location or thickness of the
body or body part being x-rayed, as in bone densitometry and
general radiography. Still another use of the disclosed process and
apparatus is for registering mammograms or other x-ray images,
where the imaged markers that can be visible or invisible, as
appropriate for a particular application, can be used as
registration marks that can be left in the registered images when
they are invisible in normal observation and this is desirable, or
can be left as visible marks, or can be removed in either case
through further processing as described above. As one non-limiting
example, the breast thickness calculated as described in this
patent specification can be displayed separately from or together
with the breast x-ray images, e.g., can be displayed to the
technician operating a system such as Selenia.TM. on a numerical
display.
[0056] Other examples of applications of the new approach include
using two of more patterns of markers, for example using two
patterns or sub-patterns as illustrated in FIG. 11, where one
pattern is above the patient's breast and another is below the
breast, such that the two patterns do not overlap or only partly
overlap in the composite x-ray image. The upper pattern can be on
the otherwise conventional compression paddle of a
mammography/tomosynthesis system and the lower pattern can be on
the top surface of an otherwise conventional magnification table
that is placed between the breast and the breast platform when it
is desired to magnify the breast x-ray image. FIG. 11 includes an
illustration to the right of a composite x-ray image that would
result from using two examples of such patterns. The markers in one
alternative embodiment would not be visible when viewing the image
in a clinical setting. In another alternative embodiment, the
imaged markers can be visible (when appropriate patterns 22 are
used) and can be removed when it is desired to view the composite
x-ray image without visible residue of imaged markers.
[0057] FIG. 12 is a flowchart illustrating one example of a process
according to the new approach disclosed in this patent
specification, and shows the following main steps of a process for
calculating the thickness of an object that is between an upper and
a lower plane and is being x-rayed. At least one pattern of
markings is present at least at one of the upper and lower planes.
In a non-limiting example, the upper plane can be at a compression
paddle and the lower surface at a breast platform in a
mammography/tomosynthesis system. The main steps can be: [0058] a.
Acquire the x-ray image, with a pattern of markers imaged within
the field of view, e.g., as illustrated in FIGS. 1 or 11; [0059] b.
Identify the approximate region of interest (ROI) where the imaged
markers are approximately expected in the x-ray image. This can be
derived through conventional calculations from knowing the
locations of the pattern(s) relative to the acquisition geometry,
including from a conventional encoder providing information
regarding the approximate height of a compression paddle; [0060] c.
Start the calculation at an estimated thickness t=t.sub.0 that is
less than the true thickness of the object. If the system reports
an approximate thickness of the object using, for example, an
encoder for paddle height, the encoder value can be used. For
example, if the encoder reports the paddle height is 5 cm, a
starting height could be 4 cm. If there is no known estimate for
the true paddle height, the starting height could be zero; [0061]
d. Generate a shift-and-added x-ray image. The amount of shifting
per image is determined from the thickness t being estimated, and
from system geometry, e.g., according to the expressions relating
thickness to known parameters discussed above in this patent
specification; [0062] e. Generate a profile by summing the image in
a direction perpendicular to the lengths of the imaged markers;
[0063] f. Calculate the one dimensional FFT of the profile; [0064]
g. Find the amplitude of the peak; [0065] h. Increment to the next
thickness to test. Typical increments depend upon the desired final
thickness measurement accuracy, and for digital
mammography/tomosynthesis can be, as a non-limiting example, in the
range of about 0.1 mm to 1 mm; [0066] i. If the thickness is less
than the desired maximum thickness to test, repeat at step d.;
[0067] j. Now find the thickness that generated the largest peak
amplitude in step g. This thickness is the calculated thickness of
the object according to the pattern above the object. Similar
processes can be used if the parameter to be calculated is the
position in space or with respect to another portion of the system,
of the upper or lower patterns of markings, or the difference in
position between the two patterns of markings, or if a parameter is
sought that is related to the position in space or in relation to a
system component, or each or two or more patterns of markings that
are approximately at the same level, such as two or more patterns
on a compression paddle that might tilt or deform in the course of
compression.
[0068] The process described immediately above can be adapted to
the more general case where it is desired to calculate the position
in space of one or more patterns of markings that are between an
x-ray source and an image plane of an x-ray imaging receptor.
[0069] FIG. 14 illustrates an upper pattern of markings "a" that is
similar to the pattern in FIG. 5a and to a sub-pattern of FIG. 5c,
and a lower pattern of markings "b" consisting of the lines of
pattern "a" broken down into four rows each consisting of much
shorter lines. Because any arbitrary pattern whose general shape of
certain other properties are known can be detected in the x-ray
image through spatial domain correlation processes (without FFT) as
long as the imaged pattern has enough signal, the arbitrary pattern
can be found with sufficient accuracy according to the approach
disclosed in this patent specification. Even a random or
pseudo-random pattern of small areas in the composite x-ray image
can be detected if it has properties that generally distinguish it
from the image of a breast or other object in the composite x-ray
image. One example of such characteristics can be edges that are
too sharp and appear too frequently to belong to an image of a
breast or another object. In the example of pattern "b" in FIG. 14,
The x-ray imaged rows of shorter lines of pattern "b" can be
shifted relative to each other through computer processing in a
direction perpendicular to their lengths to conceptually collapse
them into pattern "a." As long as pattern "b" has approximately the
same number of pixels as pattern "a," it can be found nearly
equally well in the composite x-ray image. Pattern "b" need not
consist of short lines; other shapes can also work. A pattern such
as "b" can ensure invisibility in the composite x-ray image more
reliably and under more relaxed viewing conditions.
[0070] FIG. 15 schematically illustrates measuring not only the
height of a compression paddle such as paddle 2 of FIG. 9 but also
tilt of the paddle as it compresses the patient's breast against
breast platform 12. As illustrated, support 6 for paddle 2 is
coupled with linear and angular encoders 90 in addition to being
supported for vertical motion along column 7. The linear encoder
can be an optical encoder or an encoder as used in the Selenia.TM.
system available from the common assignee. The angular encoder can
be any suitable optical encoder or a resolver that can measure the
angle of paddle 2 relevant to a reference as the paddle compresses
the patient's breast and tilts from a plane parallel to breast
platform. Both the linear and angular encoders provide digital
information to control 108 and image processor 110, which
information can be used as discussed above to assist in identifying
regions of interest to which correlation processing can be applied.
A linear encoder of this type has been in public use in this
country for more than a year. An angular encoder for breast
compression paddle that provides digital information to a control
such as 108 has not been in public use but a mammography system
with a built in potentiometer that is not providing digital
information and is not used to measure tilt and provide it for use
in the system has been in systems shown in at least one trade show
in this country.
[0071] Software that controls the process described above can be
stored in a tangible computer readable storage medium to be used as
a computer program product and/or can be transmitted via a computer
network or other transmission medium.
[0072] The above specific embodiments are illustrative, and many
variations can be introduced on these embodiments without departing
from the spirit of the disclosure or from the scope of the appended
claims. For example, elements and/or features of different examples
and illustrative embodiments may be combined with each other and/or
substituted for each other within the scope of this disclosure and
appended claims. As one of many examples, the processes disclosed
and claimed in this patent specification can be applied to CR based
x-ray imaging, where the charge storage phosphor or x-ray detection
material is not rigidly fixed in relative to the breast platform,
or similarly to digitized film images.
* * * * *
References