Method and system for the computerized assessment of breast cancer risk

• US 6,282,305 B1
• Filed: 06/05/1998
• Issued: 08/28/2001
• Est. Priority Date: 06/05/1998
• Status: Expired due to Term

• Current Assignee: ARCH Development Corporation (University of Chicago)
• Original Assignee: ARCH Development Corporation (University of Chicago)
• Inventors: Huo, Zhimin, Giger, Maryellen L.
• Primary Examiner(s): Bella, Matthew C.
• Application Number:
  US09/092,004
• Time in Patent Office: 1,180 Days
• Field of Search: 382/128, 382/133, 382/156, 382/160, 600/300, 435/29, 128/923, 128/915
• US Class Current: 382/128
• CPC Class Codes: G06F 18/2415 based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A method, system and computer readable medium for the computerized assessment of breast cancer risk, wherein a digital image of a breast is obtained and at least one feature, and typically plural features, are extracted from a region of interest in the digital. The extracted features are compared with a predetermined model associating patterns of the extracted features with a risk estimate derived from corresponding feature patterns associated with a predetermined model based on gene carrier information or clinical information, or both gene carrier information and clinical information, and a risk classification index is output as a result of the comparison. Preferred features to be extracted from the digital image include 1) one or more features based on absolute values of gray levels of pixels in said region of interest, 2) one or more features based on gray-level histogram analysis of pixels in said region of interest; 3) one or more features based on Fourier analysis of pixel values in said region of interest; and 4) one or more features based on a spatial relationship among gray levels of pixels within the region of interest.

Claims

1. A method for the computerized assessment of breast cancer risk, comprising:

obtaining a digital image of a breast of a person;

determining values of plural parenchyma features of a breast region at a predetermined parenchymal location in said digital image, comprising,

determining a value of a skewness feature based on gray-level histogram analysis of pixels within said predetermined parenchymal location,

determining values of coarseness and contrast features based on a spatial relationship among gray levels of pixels within said predetermined parenchymal location, and determining a value of at least one of a balance feature based on gray-level histogram analysis of pixels in said predetermined parenchymal location, and a first moment of power spectrum feature based on Fourier analysis of pixel values within said predetermined parenchymal location;

comparing the determined values of the plural parenchyma features with a predetermined model associating values of the plural parenchyma features with a respective risk estimate; and

outputting as a result of said comparing step a risk classification index indicating the likelihood of future onset of breast cancer.

2. The method of claim 1, further comprising:

obtaining at least one of gene carrier information and clinical information of the person; and

said comparing step comprising comparing the obtained at least one of gene carrier information and clinical information and said determined values of the plural parenchyma featured with a predetermined model associating values of the at least one of gene carrier information and clinical information and values of said plural parenchyma features with a respective risk estimate.

3. The method according to claims 1 or 2, wherein:

said determining step comprises determining plural features related to at least one of an amount of dense pattern and texture of anatomic pattern in said predetermined parenchymal location.

4. The method according to claim 3, wherein:

said comparing step comprises performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on at least one of

     1) said plural determined features and

     2) clinical information pertaining to the person from which the digital image was derived, in conjunction with said predetermined model.

5. The method according to claim 3, wherein said comparing step comprises:

merging selected features into a measure related to the risk of acquiring cancer.

6. The method according to claim 3, comprising:

obtaining gene carrier information of the person; and

said comparing step comprising comparing the obtained gene carrier and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information and values of said plural parenchyma features with a respective risk estimate.

7. The method according to claim 3, comprising:

obtaining gene carrier information and clinical information of the person; and

said comparing step comprising comparing the obtained gene carrier information and clinical information and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information, values of said clinical information and values of said plural parenchyma features with a respective risk estimate.

8. The method according to claim 4, further comprising:

relating the determined features to a likelihood of an individual having a gene mutation (biomarker).

9. The method of claim 4, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

10. The method according to claim 5, wherein said merging step comprises:

applying said selected features as inputs to a trained artificial neural network.

11. The method according to claims 1 or 2, wherein said determining step comprises:

determining the value of said balance feature within said predetermined parenchymal location.

12. The method according to claim 11, wherein said determining step comprises:

determining a value of a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location.

13. The method according to claim 11, wherein said determining step comprises:

determining the value of said balance feature based on the following relationship;

value of balance=(70% CDF−

AVE)/(AVE−

30% CDF),

where AVE=average gray level in a region of interest, 70% CDF=gray level yielding 70% of the area under the histogram of the region of interest, and 30% CDF=gray level yielding 30% of the area under the histogram of the region of interest.

14. The method according to claim 13, wherein:

said comparing step comprises performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

15. The method according to claim 13, wherein:

said comparing step comprises performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

16. The method of claim 14, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

17. The method of claim 15, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

18. The method according to claims 1 or 2, wherein said determining step comprises:

determining the value of said first moment of power spectrum feature within said predetermined parenchymal location.

19. The method according to claim 18, wherein said determining step comprises:

determining a value of a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location.

20. The method according to claim 19, wherein:

said comparing step comprises performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural extracted features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

21. The method of claim 20, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

22. The method of claim 1, wherein said determining step comprises:

determining the values of the plural parenchyma features in a central breast region behind the breast nipple in said digital image.

23. A system for the computerized assessment of breast cancer risk, comprising:

a mechanism configured to obtain a digital image of a breast of a person;

a mechanism configured to determine valued of plural parenchyma features of a breast region at a predetermined parenchymal location in said digital image, comprising,

a mechanism configured to determine a skewness feature based on gray-level histogram analysis of pixels within said predetermined parenchymal location,

a mechanism configured to determine values of coarseness and contrast features based on a spatial relationship among gray levels of pixels within said predetermined parenchymal location, and

a mechanism configured to determine a value of at least one of a balance feature based on gray-level histogram analysis of pixels in said predetermined parenchymal location,

and a first moment of power spectrum feature based on Fourier analysis of pixel values within said predetermined parenchymal location;

a mechanism configured to compare the determined values of the parenchyma features with a predetermined model associating values of the plural parenchyma features with a respective risk estimate; and

a mechanism responsive to said compare mechanism and configured to output a risk classification index indicating the likelihood of future onset of breast cancer.

24. The system of claim 23, further comprising:

a mechanism configured to obtain at least one of gene carrier information and clinical information of the person; and

said compare mechanism configured to compare the obtained at least one of gene carrier information and clinical information and said determined valued of the plural parenchyma features with a predetermined model associating values of the at least one of gene carrier information and clinical information and values of said plural parenchyma features with a respective risk estimate.

25. The system according to claims 23 or 24, wherein:

said determine mechanism mines plural features related to at least one of an amount of dense pattern and texture of anatomic pattern in said predetermined parenchymal location.

26. The system according to claim 25, wherein:

said compare mechanism performs at least one of classifier analysis, linear regression analysis and logistic regression analysis on at least one of

     1) said plural determined features and

     2) clinical information pertaining to the person from which the digital image was derived, in conjunction with said predetermined model.

27. The system according to claim 25, wherein said compare mechanism comprises:

a mechanism configured to merge selected features into a measure related to the risk of acquiring cancer.

28. The system according to claim 25, comprising:

a mechanism configured to obtain gene carrier information of the person; and

said compare mechanism configured to compare the obtained gene carrier information and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information and values of said plural parenchyma features with a respective risk estimate.

29. The system according to claim 25, comprising:

a mechanism configured to obtain gene carrier information and clinical information of the person; and

said compare mechanism configured to compare the obtained gene carrier information and clinical information and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information, values of said clinical information and values of said plural parenchyma features with a respective risk estimate.

30. The system according to claim 25, further comprising:

a mechanism configured to relate the determined features to a likelihood of an individual having a gene mutation (biomarker).

31. The system of claim 26, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

32. The system according to claim 27, wherein said merge mechanism comprises:

a mechanism configured to apply said selected features as inputs to a trained artificial neural network.

33. The system according to claims 23 or 24, wherein said determine mechanism determines the value said balance feature within said predetermined parenchymal location.

34. The system according to claim 33, wherein said determine mechanism determines a value of a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location a feature based on Fourier analysis of pixel values in said predetermined parenchymal location.

35. The system according to claim 33, wherein said determine mechanism determines the value of said balance feature based on the following relationship:

value of balance=(70% CDF−

AVE)/(AVE−

30% CDF),

where AVE=average gray level in a region of interest, 70% CDF=gray level yielding 70% of the area under the histogram of the region of interest, and 30% CDF=gray level yielding 30% of the area under the histogram of the region of interest.

36. The system according to claim 34, wherein:

said compare mechanism performs at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

37. The system according to claim 35, wherein:

said compare mechanism performs at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

38. The system of claim 36, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

39. The system of claim 37, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

40. The system according to claim 23 or 24, wherein said determine mechanism determines the value said first moment of power spectrum feature within said predetermined parenchymal location.

41. The system according to claim 40, wherein said determine mechanism determines a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location.

42. The system according to claim 41, wherein:

said compare mechanism performs at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

43. The system of claim 42, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

44. The system of claim 23, wherein said determine mechanism is configured to determine the values of said plural parenchyma features in a central breast region behind the breast nipple in said digital image.

45. A computer readable medium storing computer instructions for computerized assessment of breast cancer risk, by performing the steps of:

obtaining a digital image of a breast of a person;

determining values of plural parenchyma features of a breast region at a predetermined parenchymal location in said digital image, comprising,

determining a value of a skewness feature based on gray-level histogram analysis of pixels within said predetermined parenchymal location,

determining values of coarseness and contrast features based on a spatial relationship among gray levels of pixels within said predetermined parenchymal location, and

determining a value of at least one of a balance feature based on gray-level histogram analysis of pixels in said predetermined parenchymal location, and a first moment of power spectrum feature based on Fourier analysis of pixel values within said predetermined parenchymal location;

comparing the determined values of the plural parenchyma features with a predetermined model associating values of the plural parenchyma features with a respective risk estimate; and

outputting as a result of said comparing step a risk classification index indicating the likelihood of future onset of breast cancer.

46. The computer readable medium of claim 45, storing further computer instructions of:

obtaining at least one of gene carrier information and clinical information of the person; and

said comparing step comprising comparing the obtained at least one of gene carrier information and clinical information and said determined values of the plural parenchyma features with a predetermined model associating values of the at least one of gene carrier information and clinical information and values of said plural parenchyma features with a respective risk estimate.

47. The computer readable medium of claims 45 or 46, wherein said determining step comprises:

determining plural features related to at least one of an amount of dense pattern and texture of anatomic pattern in said predetermined parenchymal location.

48. The computer readable medium of claim 47, wherein said comparing step comprises:

performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on at least one of

     1) said plural determined features and

     2) clinical information pertaining to the person from which the digital image was derived, in conjunction with said predetermined model.

49. The computer readable medium of claim 47, wherein said comparing step comprises:

merging selected features into a measure related to the risk of acquiring cancer.

50. The computer readable medium according to claim 49, storing further computer instructions of:

obtaining gene carrier information of the person; and

said comparing step comprising comparing the obtained gene carrier and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information and values of said plural parenchyma features with a respective risk estimate.

51. The computer readable medium of claim 49, storing further computer instructions of:

obtaining gene carrier information and clinical information of the person; and

said comparing step comprising comparing the obtained gene carrier information and clinical information and said determined values of the plural parenchyma features with a predetermined model associating values of said gene carrier information, values of said clinical information and values of said plural parenchyma features with a respective risk estimate.

52. The computer readable medium of claim 47, further comprising:

relating the determined features to a likelihood of an individual having a gene mutation (biomarker).

53. The computer readable medium of claim 48, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

54. The computer readable medium of claim 49, wherein said merging step comprises:

applying said selected features as inputs to a trained artificial neural network.

55. The computer readable medium of claim 54, wherein said determining step comprises:

determining the value of said balance feature based on the following relationship;

value of balance=(70% CDF−

AVE)/(AVE−

30% CDF),

where AVE=average gray level in a region of interest, 70% CDF gram level yielding 70% of the area under the histogram of the region of interest, and 30% CDF=gray level yielding 30% of the area under the histogram of the region of interest.

56. The computer readable medium of claim 55, wherein said comparing step comprises:

performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

57. The computer readable medium of claim 56, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

58. The computer readable medium of claims 45 or 46, wherein said determining step comprises:

determining the value of said balance feature within said predetermined parenchymal location.

59. The computer readable medium of claims 45 or 46, wherein said determining step comprises:

determining the value of said first moment of power spectrum feature within said predetermined parenchymal location.

60. The computer readable medium of claim 58, wherein said determining step comprises:

determining a value of a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location.

61. The computer readable medium of claim 59, wherein said determining step comprises:

determining a value of a feature based on absolute values of gray levels of pixels in said predetermined parenchymal location.

62. The computer readable medium of claim 60, wherein said comparing step comprises:

performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

63. The computer readable medium of claim 61, wherein said comparing step comprises:

performing at least one of classifier analysis, linear regression analysis and logistic regression analysis on said plural determined features and clinical information pertaining to the person from which the digital image was derived, including the person'"'"'s age, in conjunction with said predetermined model.

64. The computer readable medium of claim 62, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

65. The computer readable medium of claim 63, wherein said classifier analysis comprises at least one of one of linear discriminant analysis and an artificial neural network.

66. The computer readable medium of claim 45, wherein said determining step comprises:

determining the value of the plural parenchyma features in a central breast region behind the breast nipple in said digital image.