Method, system, and computer software product for feature-based correlation of lesions from multiple images

• US 7,298,881 B2
• Filed: 02/14/2005
• Issued: 11/20/2007
• Est. Priority Date: 02/13/2004
• Status: Active Grant

• Current Assignee: University of Chicago
• Original Assignee: University of Chicago
• Inventors: Giger, Maryellen L, Lan, Li, Wen, HuiHua
• Primary Examiner(s): Desire; Gregory M
• Application Number:
  US11/056,368
• Publication Number:
  US 20060004278A1
• Time in Patent Office: 1,009 Days
• Field of Search: 128/922, 345/32, 378/19, 378/21, 378/37, 378/62, 382/128, 382/130, 382/131, 382/132, 382/219, 600/407, 600/425
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30068 Mammography; Breast
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/33 using feature-based methods
• CPC Class Codes: G06T 7/35 using statistical methods
• CPC Class Codes: G06T 7/97 Determining parameters from multiple pictures depth or shape recovery from multiple images G06T7/55; stereo camera calibration G06T7/85
• CPC Class Codes: Y10S 128/922 including image analysis

Abstract

A method, system, and computer software product for correlating medical images, comprising: obtaining first image data representative of a first medical image including a first abnormality; obtaining second image data representative of a second medical image including a second abnormality; determining at least one feature value for each of the first and second abnormalities using the first and second image data; calculating, based on the determined feature values, a likelihood value indicative of a likelihood that the first and second abnormalities are a same abnormality; and outputting the determined likelihood value.

Claims

1. A computer-implemented method for correlating medical images, comprising:

obtaining first image data representative of a first medical image including a first abnormality;

obtaining second image data representative of a second medical image including a second abnormality;

determining at least one feature value for each of the first and second abnormalities using the first and second image data;

calculating, based on the determined feature values, a likelihood value indicative of a likelihood that the first and second abnormalities are a same abnormality; and

outputting the determined likelihood value,

wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image derived using a first modality, the first modality being one of mammography, sonography, and magnetic resonance imaging; and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image derived using a second modality different from the first modality.

2. The method of claim 1, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first modality, in a given view;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image derived using a second modality, in a view different from the given view.

3. The method of claim 1, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first modality, in a given time;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using the first modality, at a time different from the given time.

4. The method of claim 1, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first protocol of a first modality;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using a second protocol of the first modality, the second protocol being different from the first protocol.

5. The method of claim 1, wherein the determining step comprises:

automatically segmenting the first and second abnormalities.

6. The method of claim 1, wherein the determining step comprises:

identifying at least one maximally correlated feature using a canonical correlation analysis; and

determining a value for each of the at least one maximally correlated feature for each of the first and second abnormalities.

7. The method of claim 6, wherein the identifying step comprises:

selecting the at least one maximally correlated feature from a list of candidate features including filtered ARD, filtered margin sharpness, posterior acoustic behavior, texture, NRG, average gray value, contrast, and diameter.

8. The method of claim 1, wherein the determining step comprises:

evaluating a measure of similarity between maximally correlated features to determine feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

9. The method of claim 8, wherein the calculating step comprises:

using a likelihood ratio test based on the feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

10. The method of claim 1, wherein the calculating step comprises:

applying the determined feature values to a classifier to obtain the likelihood value.

11. A system for correlating medical images, comprising:

a mechanism configured to obtain first image data representative of a first medical image including a first abnormality;

a mechanism configured to obtain second image data representative of a second medical image including a second abnormality;

a mechanism configured to determine at least one feature value for each of the first and second abnormalities using the first and second image data;

a mechanism configured to calculate, based on the determined feature values, a likelihood value indicative of a likelihood that the first and second abnormalities are a same abnormality; and

a mechanism configured to output the determined likelihood value,

wherein the mechanism configured to obtain the first image data comprises a mechanism configured to obtain first image data representative of the first medical image using a first modality, the first modality being one of mammography, sonography, and magnetic resonance imaging; and

the mechanism configured to obtain the second image data comprises a mechanism configured to obtain second image data representative of the second medical image using a second modality different from the first modality.

12. The system of claim 11, wherein the mechanism configured to obtain the first image data comprises a mechanism configured to obtain first image data representative of the first medical image using a first modality, in a given view;

and

the mechanism configured to obtain the second image data comprises a mechanism configured to obtain second image data representative of the second medical image using the first modality, in a view different from the given view.

13. The system of claim 11, wherein the mechanism configured to obtain the first image data comprises a mechanism configured to obtain first image data representative of the first medical image using a first modality, at a given time;

and

the mechanism configured to obtain the second image data comprises a mechanism configured to obtain second image data representative of the second medical image using the first modality, at a time different from the given time.

14. The system of claim 11, wherein the mechanism configured to obtain the first image data comprises a mechanism configured to obtain first image data representative of the first medical image using a first protocol of a first modality;

and

the mechanism configured to obtain the second image data comprises a mechanism configured to obtain second image data representative of the second medical image using a second protocol of the first modality, the second protocol being different from the first protocol.

15. The system of claim 11, wherein the mechanism configured to determine comprises:

a mechanism configured to automatically segment the first and second abnormalities.

16. The system of claim 11, wherein the mechanism configured to determine comprises:

a mechanism configured to identify at least one maximally correlated feature use a canonical correlation analysis; and

a mechanism configured to determine a value for each of the at least one maximally correlated feature for each of the first and second abnormalities.

17. The system of claim 16, wherein the mechanism configured to identify the at least one maximally correlated feature comprises:

a mechanism configured to select the at least one maximally correlated feature from a list of candidate features including filtered ARD, filtered margin sharpness, posterior acoustic behavior, texture, NRG, average gray value, contrast, and diameter.

18. The system of claim 11, wherein the mechanism configured to determine comprises:

a mechanism configured to evaluate a measure of similarity between maximally correlated features to determine feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

19. The system of claim 18, wherein the mechanism configured to calculate comprises:

a mechanism configured to use a likelihood ratio test based on the feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

20. The system of claim 11, wherein the mechanism configured to calculate comprises:

a mechanism configured to apply the determined feature values to a classifier to obtain the likelihood value.

21. A computer program product which stores, on a computer-readable medium, instructions for execution on a computer system, which when executed by the computer system, causes the computer system to perform the steps of:

obtaining first image data representative of a first medical image including a first abnormality;

obtaining second image data representative of a second medical image including a second abnormality;

determining at least one feature value for each of the first and second abnormalities using the first and second image data;

calculating, based on the determined feature values, a likelihood value indicative of a likelihood that the first and second abnormalities are a same abnormality; and

outputting the determined likelihood values,

wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first modality, the first modality being one of mammography, sonography, and magnetic resonance imaging; and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using a second modality different from the first modality.

22. The computer program product of claim 21, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first modality, in a given view;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using the first modality, in a view different from the given view.

23. The computer program product of claim 21, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first modality, at a given time;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using the first modality, at a time different from the given time.

24. The computer program product of claim 21, wherein the step of obtaining the first image data comprises obtaining first image data representative of the first medical image using a first protocol of a first modality;

and

the step of obtaining the second image data comprises obtaining second image data representative of the second medical image using a second protocol of the first modality, the second protocol being different from the first protocol.

25. The computer program product of claim 21, wherein the determining step comprises:

automatically segmenting the first and second abnormalities.

26. The computer program product of claim 21, wherein the determining step comprises:

identifying at least one maximally correlated feature using a canonical correlation analysis; and

determining a value for each of the at least one maximally correlated feature for each of the first and second abnormalities.

27. The computer program product of claim 26, wherein the identifying step comprises:

selecting the at least one maximally correlated feature from a list of candidate features including filtered ARD, filtered margin sharpness, posterior acoustic behavior, texture, NRG, average gray value, contrast, and diameter.

28. The computer program product of claim 21, wherein the determining step comprises:

evaluating a measure of similarity between maximally correlated features to determine feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

29. The computer program product of claim 28, wherein the calculating step comprises:

using a likelihood ratio test based on the feature-conditioned likelihoods that the first and second abnormalities are the same abnormality.

30. The computer program product of claim 21, wherein the calculating step comprises:

applying the determined feature values to a classifier to obtain the likelihood value.