Automated method and system for the detection and classification of abnormal lesions and parenchymal distortions in digital medical images
First Claim
1. A method for automated analysis of abnormalities in the form of lesions and parenchymal distortions using digital images, comprising:
- generating at least first and second image data from respective of at least first and second digital images derived from at least one selected portion of an object; and
correlating said at least first and second image data to produce correlated data in which normal anatomical structured background is removed.
2 Assignments
0 Petitions
Accused Products
Abstract
A method for automated analysis of abnormalities in the form of lesions and parenchymal distortions using digital images, including generating image data from respective of digital images derived from at least one selected portion of an object, for example, from mammographical digital images of the left and right breasts. The image data from each of the digital images are then correlated to produce correlated data in which normal anatomical structured background is removed. The correlated data is then searched using one or more predetermined criteria to identify in at least one of the digital images an abnormal region represented by a portion of the correlated data which meets the predetermined criteria. The location of the abnormal region is then indicated, and the indicated location is then subjected to classification processing to determine whether or not the abnormal region is benign or malignant. Classification is performed based on the degree of spiculations of the identified abnormal region. In order to enhance the process of searching for abnormal regions, in one embodiment the gray-level frequency-distributions of two or more images are matched by matching the cumulative gray-level histograms of the images in question.
253 Citations
108 Claims
-
1. A method for automated analysis of abnormalities in the form of lesions and parenchymal distortions using digital images, comprising:
-
generating at least first and second image data from respective of at least first and second digital images derived from at least one selected portion of an object; and correlating said at least first and second image data to produce correlated data in which normal anatomical structured background is removed. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45)
-
2. The method according to claim 1, further comprising:
-
searching said correlated data using at least one predetermined criterion to identify in at least one of said digital images an abnormal region represented by a portion of said correlated data which meets said predetermined criterion; and indicating the location of said abnormal region in said at least one of said digital images.
-
-
3. The method according to claim 2, wherein said generating step comprises:
-
determining at least first and second histograms of respective of said at least first and second digital images, determining at least first and second cumulative histograms of said at least first and second histograms, respectively, and matching at least the second cumulative histogram to the first cumulative histogram by modifying data values of pixels of at least the second digital image based on a predetermined relationship between the data of the at least first and second cumulative histograms, said second image data corresponding to modified pixel values of said second digital image and said first image data corresponding to pixel values of the first digital image.
-
-
4. The method according to claim 3, wherein said generating step comprises:
determining N at least first and second threshold values, and determining N at least first and second threshold images for each of said at least first and second digital images by comparing each pixel of said at least first and second digital images with each of respective of said N at least first and second threshold values and assigning each said pixel a first predetermined value when said pixel is above the threshold value and a second predetermined value when said pixel is below the threshold value, said first image data being N image data corresponding to the N first threshold images and said second image data being N image data corresponding to the N second threshold images.
-
5. The method according to claim 4, wherein said step of determining said N at least first and second threshold images comprises:
assigning to each said pixel of said at least first and second digital images a predetermined first constant value or a predetermined second constant value when that pixel has a value above or below said threshold value, respectively.
-
6. The method according to claim 4, wherein said step of determining said N at least first and second threshold images comprises:
-
assigning each said pixel of said at least first and second digital images the value of that pixel in the respective digital image when that pixel has a value above said threshold value, and assigning each said pixel of said at least first and second digital images a predetermined constant value when that pixel has a value below said threshold value.
-
-
7. The method according to claim 4, wherein said step of determining said N at least first and second threshold values comprises:
-
determining at least first and second histograms of said at least first and second digital images, respectively, and defining said N at least first and second threshold values as being the pixel values at selected percentages of said at least first and second histograms, respectively.
-
-
8. The method according to claim 1, wherein said correlating step comprises:
forming a difference image based on the difference between said first and second image data.
-
9. The method according to claim 3, wherein said correlating step comprises:
forming a difference image based on the difference between said first and second image data.
-
10. The method according to claim 4, wherein said correlating step comprises:
forming N difference images based on the difference between the N first threshold images and respective of the N second threshold images.
-
11. The method according to claim 10, wherein said correlating step further comprises:
-
forming a first runlength image in which each pixel of the first runlength image corresponds to an identical pixel location in said N difference images and is assigned a value in dependence on the largest number of said identical pixels which in consecutive order of said N difference images have a predetermined positive value, and forming a second runlength image in which each pixel of the second runlength image corresponds to an identical pixel location in said N images and is assigned a value in dependence on the largest number of said identical pixels which in consecutive order of said N difference images have a predetermined negative value.
-
-
12. The method according to claim 9, wherein said searching step comprises:
-
comparing the value of each pixel of said difference image with a predetermined threshold value, and identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold value.
-
-
13. The method according to claim 12, wherein said searching step comprises:
determining the size of each identified abnormal region and further identifying as an abnormal region only those previously identified possible abnormal regions which also exceed a predetermined size.
-
14. The method according to claim 12, wherein said searching step comprises:
-
determining which of said identified abnormal regions are in a boundary region of said difference image, and further identifying as abnormal regions only those previously identified possible abnormal regions also lying outside said boundary region.
-
-
15. The method according to claim 12, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
16. The method according to claim 13, wherein said searching step comprises:
-
determining which of said identified abnormal regions are in a boundary region of said difference image, and further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
17. The method according to claim 13, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
18. The method according to claim 14, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital views, and said searching step comprises:
-
determining which of said identified abnormal region derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
19. The method according to claim 16, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as an abnormal region only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
20. The method according to claim 11, wherein said searching step comprises:
-
comparing the value of each pixel of said first runlength difference image with a predetermined threshold value, and identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold value.
-
-
21. The method according to claim 20, wherein said searching step comprises:
determining the size of each identified abnormal region and further identifying as an abnormal region only those previously identified abnormal regions which exceed a predetermined size.
-
22. The method according to claim 20, wherein said searching step comprises:
-
determining which of said identified abnormal regions are in a boundary region of said first runlength image, and further identifying as abnormal regions only those previously identified abnormal regions lying outside said boundary region.
-
-
23. The method according to claim 20, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
24. The method according to claim 21, wherein said searching step comprises:
-
determining which of said abnormal regions are in a boundary region of said first runlength image, and further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
25. The method according to claim 21, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
26. The method according to claim 22, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal region derived from said third and fourth digital images.
-
-
27. The method according to claim 24, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
28. The method according to claim 11, wherein said searching step comprises:
-
comparing the value of each pixel of said second runlength image with a predetermined threshold value, and identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold.
-
-
29. The method according to claim 28, wherein said searching step comprises:
determining the size of each identified abnormal region and further identifying as a abnormal region only those previously identified abnormal regions which exceed a predetermined size.
-
30. The method according to claim 28, wherein said searching step comprises:
determining which of said identified abnormal regions are in a boundary region of said second runlength image, and further identifying as abnormal regions only those previously identified abnormal regions lying outside said boundary region.
-
31. The method according to claim 28, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images a representing different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
32. The method according to claim 29, wherein said searching step comprises:
-
determining which of said abnormal regions are in a boundary region of said second runlength image, and further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
33. The method according to claim 29, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images a representing different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
34. The method according to claim 30, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
35. The method according to claim 32, wherein said generating, correlating and searching steps are repeated to identify abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images, and said searching step comprises:
-
determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
36. The method according to claim 1, comprising:
aligning said digital images with respect to each other prior to performing said correlating step.
-
37. The method according to claim 2, wherein said indicating step comprises:
indicating a center of said abnormal region.
-
38. The method according to claim 2, further comprising:
-
classifying said abnormal region as being malignant or benign; and said indicating step comprising indicating a result of said classifying step.
-
-
39. The method according to claim 2, comprising:
-
classifying said abnormal region as being malignant or benign, including determining a degree of spiculations of said abnormal region, and characterizing said abnormal region as being malignant or benign based on the determined degree of spiculations.
-
-
40. The method according to claim 39, wherein said step of determining a degree of spiculation of said abnormal region comprises:
-
determining a center of said abnormal region, determining in at least one of said first and second digital images a range of amplitude values encompassing said abnormal region, and determining all contiguous pixels surrounding and contiguous with the center of said abnormal region and having a value within said range of amplitude values, said contiguous pixels thereby defining a grown image of said abnormal region.
-
-
41. The method according to claim 40, further comprising:
determining the border of said grown image.
-
42. The method according to claim 39, wherein said classifying step further comprises:
-
determining a border of said abnormal region, smoothing spatial locations of said border of said abnormal region to produce a smoothed border, determining a difference between said border of said abnormal region and said smoothed border, determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said smoothed border, and characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and first moment of power spectrum.
-
-
43. The method according to claim 39, wherein said classifying step comprises:
-
blurring of said abnormal region to define a blurred region, determining the area of said abnormal region and the area of said blurred region, determining a difference between the area of said abnormal region and the area of said blurred region, and characterizing said abnormal region as malignant or benign based on the difference between said areas.
-
-
44. The method according to claim 39, wherein said classifying step comprises:
-
determining a border of said abnormal region, blurring of said abnormal region by spatial filtering to define a blurred region, determining a border of said blurred region, determining a difference between said border of said abnormal region and said border of said blurred region, determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said border of said blurred region, and characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
45. The method according to claim 39, wherein said classifying step comprises:
-
blurring of said abnormal region by spatial filtering to define a blurred region, determining a first discrete Fourier transform of said abnormal region and a second discrete Fourier transform of said blurred region, determining a difference between spectra of the first and second discrete Fourier transforms, determining an RMS variation and a first moment of a power spectrum of said difference between spectra, and characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
2. The method according to claim 1, further comprising:
-
-
46. A method for automated classification of an abnormal region in the form of a lesion or a parenchymal distortion in a digital image of an object, comprising:
-
determining a degree of spiculations in said abnormal region; and characterizing said abnormal region as being malignant or benign based on the determined degree of spiculations. - View Dependent Claims (47, 48, 49, 50, 51, 52)
-
47. The method according to claim 46, wherein said step of determining a degree of spiculations of said abnormal region comprises:
-
determining a center of said abnormal region, determining in said digital image a range of amplitude values encompassing said abnormal region, and determining all contiguous pixels surrounding and contiguous with the center of said abnormal region and having a value within said range of amplitude values, said contiguous pixels thereby defining a grown image of said abnormal region.
-
-
48. The method according to claim 47, further comprising:
determining the border of said grown image.
-
49. The method according to claim 46, wherein:
-
said step of determining a degree of spiculations comprises, determining a border of said abnormal region, smoothing spatial locations of said border of said abnormal region to produce a smoothed border, determining a difference between said border of said abnormal region and said smoothed border, and determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said smoothed border; and said characterizing step comprises, characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and first moment of power spectrum.
-
-
50. The method according to claim 46, wherein:
-
said step of determining a degree of spiculations comprises, blurring of said abnormal region to define a blurred region, determining the area of said abnormal region and the area of said blurred region, and determining a difference between the area of said abnormal region and the area of said blurred region; and said characterizing step comprises, characterizing said abnormal region as malignant or benign based on the difference between said areas.
-
-
51. The method according to claim 46, wherein:
-
said step of determining a degree of spiculations comprises, determining a border of said abnormal region, blurring of said abnormal region by spatial filtering to define a blurred region, determining a border of said blurred region, determining a difference between said border of said abnormal region and said border of said blurred region, and determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said border of said blurred region; and said characterizing step comprises, characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
52. The method according to claim 46, wherein said step of determining a degree of spiculations comprises,
blurring of said abnormal region by spatial filtering to define a blurred region, determining a first discrete Fourier transform of said abnormal region and a second discrete Fourier transform of said blurred region, determining a difference between spectra of the first and second discrete Fourier transforms, and determining an RMS variation and first moment of a power spectrum of said difference between spectra; - and
said characterizing steps comprises, characterizing said abnormal region as malignant or benign base on the relationship of the RMS variation and the first moment of power spectrum.
- and
-
47. The method according to claim 46, wherein said step of determining a degree of spiculations of said abnormal region comprises:
-
-
53. A method of processing at least first and second digital images represented by at least first and second digital data, respectively, of at least one selected portion of an object, comprising:
-
determining at least first and second histograms of respective of said at least first and second digital images; determining at least first and second cumulative histograms of said at least first and second histograms, respectively; and matching at least the second cumulative histogram to the first cumulative histogram by modifying at least said second digital data based on a predetermined relationship between the data of the a least first and second cumulative histograms. - View Dependent Claims (54)
-
54. The method according to claim 53, wherein said at least first and second cumulative histograms each have a first axis defining data values and a second axis defining a total percent of pixels having data values equal to or less than a respective data value on said first axis, and said matching step comprises:
-
determining for each data value on the first axis of said second cumulative histogram a respective percent of pixels on the second axis of said second cumulative histogram, determining for each respective percent determined in the preceding step a respective data value corresponding thereto on the first axis of the first cumulative histogram, and converting the data value of each pixel having a data value at a particular percent in said second cumulative histogram to the respective data value at the same particular percent in said first cumulative histogram determined in said preceding determining step.
-
-
54. The method according to claim 53, wherein said at least first and second cumulative histograms each have a first axis defining data values and a second axis defining a total percent of pixels having data values equal to or less than a respective data value on said first axis, and said matching step comprises:
-
-
55. A system for automated analysis of abnormalities in the form of lesions and parenchymal distortions using digital images, comprising:
-
means for generating at least first and second image data from respective of at least first and second digital images derived from at least one selected portion of an object; and means for correlating said at least first and second image data to produce correlated data in which normal anatomical structured background is removed. - View Dependent Claims (56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99)
-
56. The system according to claim 55, further comprising:
-
means for searching said correlated data using at least one predetermined criterion to identify in at least one of said digital images an abnormal region represented by a portion of said correlated data which meets said predetermined criterion; and means for indicating the location of said abnormal region in said at least one of said digital images.
-
-
57. The system according to claim 56, wherein said means for generating comprises:
-
means for determining at least first and second histograms of respective of said at least first and second digital images, means for determining at least first and second cumulative histograms of said at least first and second histograms, respectively, and means for matching at least the second cumulative histogram to the first cumulative histogram by modifying data values of pixels of at least the second digital image based on a predetermined relationship between the data of the at least first and second cumulative histograms, said second image data corresponding to modified pixel values of said second digital image and said first image data corresponding to pixel values of the first digital image.
-
-
58. The system according to claim 57, wherein said generating comprises:
-
means for determining N at least first and second threshold values, and means for determining N at least first and second threshold images for each of said at least first and second digital images by comparing each pixel of said at least first and second digital images with each of respective of said N at least first and second threshold values and assigning each said pixel a first predetermined value when said pixel is above the threshold value and a second predetermined value when said pixel is below the threshold value, said first image data being N image data corresponding to the N first threshold images and said second image data being N image data corresponding to the N second threshold images.
-
-
59. The system according to claim 58, wherein said means for determining said N at least first and second threshold images comprises:
means for assigning to each said pixel of said at least first and second digital images a predetermined first constant value or a predetermined second constant value when that pixel has a value above or below said threshold value, respectively.
-
60. The system according to claim 58, wherein said means for determining said N at least first and second threshold images comprises:
-
means for assigning each said pixel of said at least first and second digital images the value of that pixel in the respective digital image when that pixel has a value above said threshold value, and means for assigning each said pixel of said at least first and second digital images a predetermined constant value when that pixel has a value below said threshold value.
-
-
61. The system according to claim 58, wherein said means for determining said N at least first and second threshold values comprises:
-
means for determining at least first and second histograms of said at least first and second digital images, respectively, and means for defining said N at least first and second threshold values as being the pixel values at selected percentages of said at least first and second histograms, respectively.
-
-
62. The system according to claim 55, wherein said means for correlating comprises:
means for forming a difference image based on the difference between said first and second image data.
-
63. The system according to claim 57, wherein said means for correlating comprises:
means for forming a difference image based on the difference between said first and second image data.
-
64. The system according to claim 58, wherein said means for correlating comprises:
means for forming N difference images based on the difference between the N first threshold images and respective of the N second threshold images.
-
65. The system according to claim 64, wherein said means for correlating comprises:
-
means for forming a first runlength image in which each pixel of the first runlength image corresponds to an identical pixel location in said N difference images and is assigned a value in dependence on the largest number of said identical pixels which in consecutive order of said N difference images have a predetermined positive value, and means for forming a second runlength image in which each pixel of the second runlength image corresponds to an identical pixel location in said N images and is assigned a value in dependence on the largest number of said identical pixels which in consecutive order of said N difference images have a predetermined negative value.
-
-
66. The system according to claim 63, wherein said means for searching comprises:
-
means for comparing the value of each pixel of said difference image with a predetermined threshold value, and means for identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold value.
-
-
67. The system according to claim 66, wherein said means for searching comprises:
means for determining the size of each identified abnormal region and further identifying as an abnormal region only those previously identified possible abnormal regions which also exceed a predetermined size.
-
68. The system according to claim 66, wherein said means for searching comprises:
-
means for determining which of said identified abnormal regions are in a boundary region of said difference image, and means for further identifying as abnormal regions only those previously identified possible abnormal regions also lying outside said boundary region.
-
-
69. The system according to claim 66, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
70. The system according to claim 67, wherein said means for searching comprises:
-
means for determining which of said identified abnormal regions are in a boundary region of said difference image, and means for further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
71. The system according to claim 67, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
72. The system according to claim 68, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital views are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
73. The system according to claim 70, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as an abnormal region only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
74. The system according to claim 68, wherein said means for searching comprises:
-
means for comparing the value of each pixel of said first runlength difference image with a predetermined threshold value, and means for identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold value.
-
-
75. The system according to claim 74, wherein said means for searching comprises:
means for determining the size of each identified abnormal region and further identifying as an abnormal region only those previously identified abnormal regions which exceed a predetermined size.
-
76. The system according to claim 74, wherein said means for searching comprises:
-
means for determining which of said identified abnormal regions are in a boundary region of said first runlength image, and means for further identifying as abnormal regions only those previously identified abnormal regions lying outside said boundary region.
-
-
77. The system according to claim 74, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
78. The system according to claim 75, wherein said means for searching comprises:
-
means for determining which of said abnormal regions are in a boundary region of said first runlength image, and means for further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
79. The system according to claim 75, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
80. The system according to claim 76, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal region derived from said third and fourth digital images.
-
-
81. The system according to claim 78, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
82. The system according to claim 65, wherein said means for searching comprises:
-
means for comparing the value of each pixel of said second runlength image with a predetermined threshold value, and means for identifying as an abnormal region each region formed by contiguous of those pixels which have values exceeding the predetermined threshold.
-
-
83. The system according to claim 82, wherein said means for searching comprises:
means for determining the size of each identified abnormal region and further identifying as a abnormal region only those previously identified abnormal regions which exceed a predetermined size.
-
84. The system according to claim 82, wherein said means for searching comprises:
-
determining which of said identified abnormal regions are in a boundary region of said second runlength image, and means for further identifying as abnormal regions only those previously identified abnormal regions lying outside said boundary region.
-
-
85. The system according to claim 82, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
86. The system according to claim 83, wherein said means for searching comprises:
-
means for determining which of said abnormal regions are in a boundary region of said second runlength image, and means for further identifying as abnormal regions only those previously identified abnormal regions also lying outside said boundary region.
-
-
87. The system according to claim 83, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
88. The system according to claim 84, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
89. The system according to claim 86, wherein abnormal regions of said object derived from third and fourth digital images representing a different view of said object relative to said first and second digital images are identified, and said means for searching further comprises:
-
means for determining which of said identified abnormal regions derived from said first and second digital images have spatial correspondence with identified abnormal regions derived from said third and fourth digital images, and means for further identifying as abnormal regions only those previously identified abnormal regions derived from said first and second digital images which also have spatial correspondence with previously identified abnormal regions derived from said third and fourth digital images.
-
-
90. The system according to claim 55, comprising:
means for aligning said digital images with respect to each other.
-
91. The system according to claim 56, wherein said means for indicating comprises:
means for indicating a center of said abnormal region.
-
92. The system according to claim 57, further comprising:
-
means for classifying said abnormal region as being malignant or benign; and said means for indicating comprising means for indicating a result of said classifying step
-
-
93. The system according to claim 56, comprising:
-
means for classifying said abnormal region as being malignant or benign, including means for determining a degree of spiculations of said abnormal region, and means for characterizing said abnormal region as being malignant or benign based on the determined degree of spiculations.
-
-
94. The system according to claim 93, wherein said means for determining a degree of spiculation of said abnormal region comprises:
-
means for determining a center of said abnormal region, means for determining in at least one of said first and second digital images a range of amplitude values encompassing said abnormal region, and means for determining all contiguous pixels surrounding and contiguous with the center of said abnormal region and having a value within said range of amplitude values, said contiguous pixels thereby defining a grown image of said abnormal region.
-
-
95. The system according to claim 94, further comprising:
means for determining the border of said grown image.
-
96. The system according to claim 93, wherein said means for classifying further comprises:
-
means for determining a border of said abnormal region, means for smoothing spatial locations of said border of said abnormal region to produce a smoothed border, means for determining a difference between said border of said abnormal region and said smoothed border, means for determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said smoothed border, and means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and first moment of power spectrum.
-
-
97. The system according to claim 93, wherein said means for classifying comprises:
-
means for blurring of said abnormal region to define a blurred region, means for determining the area of said abnormal region and the area of said blurred region, means for determining a difference between the area of said abnormal region and the area of said blurred region, and means for characterizing said abnormal region as malignant or benign based on the difference between said areas.
-
-
98. The system according to claim 93, wherein said means for classifying comprises:
-
means for determining a border of said abnormal region, means for blurring of said abnormal region by spatial filtering to define a blurred region, means for determining a border of said blurred region, means for determining a difference between said border of said abnormal region and said border of said blurred region, means for determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said border of said blurred region, and means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
99. The system according to claim 93, wherein said means for classifying comprises:
-
means for blurring of said abnormal region by spatial filtering to define a blurred region, means for determining a first discrete Fourier transform of said abnormal region and a second discrete Fourier transform of said blurred region, means for determining a difference between spectra of the first and second discrete Fourier transforms, means for determining an RMS variation and a first moment of a power spectrum of said difference between spectra, and means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
56. The system according to claim 55, further comprising:
-
-
100. A system for automated classification of an abnormal region in the form of a lesion or a parenchymal distortion in a digital image of an object, comprising:
-
means for determining a degree of spiculations in said abnormal region; and means for characterizing said abnormal region as being malignant or benign based on the determined degree of spiculations. - View Dependent Claims (101, 102, 103, 104, 105, 106)
-
101. The system according to claim 100, wherein said means for determining a degree of spiculations of said abnormal region comprises:
-
means for determining a center of said abnormal region, means for determining in said digital image a range of amplitude values encompassing said abnormal region, and means for determining all contiguous pixels surrounding and contiguous with the center of said abnormal region and having a value within said range of amplitude values, said contiguous pixels thereby defining a grown image of said abnormal region.
-
-
102. The system according to claim 101, further comprising:
means for determining the border of said grown image.
-
103. The system according to claim 100, wherein:
-
said means for determining a degree of spiculations comprises, means for determining a border of said abnormal region, means for smoothing spatial locations of said border of said abnormal region to produce a smoothed border, and means for determining a difference between said border of said abnormal region and said smoothed border, and means for determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said smoothed border; and said means for characterizing comprises, means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and first moment of power spectrum.
-
-
104. The system according to claim 100, wherein:
-
said means for determining a degree of spiculations comprises, means for blurring of said abnormal region to define a blurred region, means for determining the area of said abnormal region and the area of said blurred region, and means for determining a difference between the area of said abnormal region and the area of said blurred region; and said means for characterizing comprises, means for characterizing said abnormal region as malignant or benign based on the difference between said areas.
-
-
105. The system according to claim 100, wherein:
-
said means for determining a degree of spiculations comprises, means for determining a border of said abnormal region, means for blurring of said abnormal region by spatial filtering to define a blurred region, means for determining a border of said blurred region, means for determining a difference between said border of said abnormal region and said border of said blurred region, and means for determining an RMS variation and a first moment of power spectrum of the difference between said border of said abnormal region and said border of said blurred region; and said means for characterizing comprises, means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
-
-
106. The system according to claim 100, wherein said means for determining a degree of spiculations comprises,
means for blurring of said abnormal region by spatial filtering to define a blurred region, means for determining a first discrete Fourier transform of said abnormal region and a second discrete Fourier transform of said blurred region, means for determining a difference between spectra of the first and second discrete Fourier transforms, and means for determining an RMS variation and first moment of a power spectrum of said difference between spectra; - and
said means for characterizing comprises, means for characterizing said abnormal region as malignant or benign based on the relationship of the RMS variation and the first moment of power spectrum.
- and
-
101. The system according to claim 100, wherein said means for determining a degree of spiculations of said abnormal region comprises:
-
-
107. A system for processing at least first and second digital images represented by at least first and second digital data, respectively, of at least one selected portion of an object, comprising:
-
means for determining at least first and second histograms of respective of said at least first and second digital images; means for determining at least first and second cumulative histograms of said at least first and second histograms, respectively; and means for matching at least he second cumulative histogram to the first cumulative histogram by modifying at least said second digital data based on a predetermined relationship between the data of the at least first and second cumulative histograms. - View Dependent Claims (108)
-
108. The system according to claim 107, wherein said at least first and second cumulative histograms each have a first axis defining data values and a second axis defining a total percent of pixels having data values equal to or less than a respective data value on said first axis, and said means for matching comprises:
-
first means for determining for each data value on the first axis of said second cumulative histogram a respective percent of pixels on the second axis of said second cumulative histogram, second means for determining for each respective percent determined by said first means a respective data value corresponding thereto on the first axis of the first cumulative histogram, and means for converting the data value of each pixel having a data value at a particular percent in said second cumulative histogram to the respective data value at the same particular percent in said first cumulative histogram determined by said second determining means.
-
-
108. The system according to claim 107, wherein said at least first and second cumulative histograms each have a first axis defining data values and a second axis defining a total percent of pixels having data values equal to or less than a respective data value on said first axis, and said means for matching comprises:
-
Specification
- Resources
-
Current AssigneeARCH Development Corporation (University of Chicago)
-
Original AssigneeARCH Development Corporation (University of Chicago)
-
InventorsYin, Fang-Fang, Metz, Charles E., Giger, Maryellen L., Doi, Kunio
-
Primary Examiner(s)Boudreau, Leo H.
-
Application NumberUS07/383,097Time in Patent Office1,096 DaysField of Search382/6, 382/18, 382/51, 364/413.19, 364/413.17, 364/413.23, 364/413.13US Class Current382/128CPC Class CodesA61B 6/502 for diagnosis of breast, i....G06T 2207/30068 Mammography; BreastG06T 7/0012 Biomedical image inspection