System and method for helping to determine the condition of tissue

• US 6,757,412 B1
• Filed: 10/19/1999
• Issued: 06/29/2004
• Est. Priority Date: 10/21/1998
• Status: Expired due to Fees

• Current Assignee: Computerized Thermanl Imaging Company
• Original Assignee: Computerized Thermanl Imaging Company
• Inventors: Waagen, Donald E., Parsons, Mark David, Rich, David Cardon, Friefer, Cliff Eugene, Cothren, Robert Mack, Birbeck, John Scott
• Primary Examiner(s): Ahmed, Samir
• Assistant Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US09/425,042
• Time in Patent Office: 1,715 Days
• Field of Search: 382/125, 382/128, 382/115, 382/117, 382/118, 600/306, 600/473-475, 600/310, 378/141, 250/316.1, 250/338.1, 356/51, 356/71, 374/10, 374/121, 374/124, 348/77, 348/78, 348/161, 340/825.36, 340/825.14, 128/922, 377/25, 702/130-136
• US Class Current: 382/128
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: Y10S 128/922 including image analysis

Abstract

A system and method is disclosed for helping determine a condition of selected tissue of a patient based on a set of thermal images of the selected tissue acquired over a time period. The image set has image regions with corresponding characteristics indicative of a temperature condition of an associated part of the selected tissue, which had been subjected to a thermal challenge during part of the image acquisition time period. Feature values are determined based on the characteristics of selected image regions. A classifier is applied to the features to provide an indication of the condition of the selected tissue, such as indicating a likelihood of malignancy.

Claims

1. A system for helping to determine a condition of selected tissue of a patient, said system comprising:

memory storing a plurality of images of the selected tissue acquired over a time period to define an image set, each of said images of said image set including a plurality of image regions, each of said image regions having a characteristic that is variable over the time period indicative of a temperature condition of an associated part of the selected tissue, said memory also storing a statistical classifier that describes temperature characteristics of previously diagnosed tissue samples having a known tissue condition; and

a processor connected with said memory, said processor being operative to determine at least one feature value for at least one of said image regions, said at least one feature value being indicative of at least part of the characteristic of said at least one image region over the time period, said processor being operative to apply said statistical classifier to said at least one feature value to provide an indication of the condition of the part of the selected tissue of the patient associated with said at least one image region.

2. A system as set forth in claim 1 wherein said classifier describes a statistical distribution of a plurality of predetermined feature values, said predetermined feature values being derived from image sets of previously diagnosed tissue having a known tissue condition, said processor applying said classifier to said at least one feature value to determine a value for the part of the selected tissue associated with said at least one image region indicating a likelihood that the part of the selected tissue associated with said at least one image region belongs to a class of tissue described by said classifier.

3. A system as set forth in claim 2 wherein said classifier has a first part describing a distribution of predetermined feature values derived from previously diagnosed tissue having a first tissue condition, said classifier having a second part describing another distribution of predetermined feature values derived from previously diagnosed tissue having a second tissue condition which is different from the first tissue condition, said processor determining (i) a likelihood that the part of the selected tissue associated with said at least one image region has the first tissue condition and (ii) a likelihood that the part of the selected tissue associated with said at least one image region has the second tissue condition.

4. A system as set forth in claim 3 wherein the first tissue condition is malignant and the second tissue condition is benign, said processor applying said classifier to said at least one feature to indicate a likelihood that the part of the selected tissue associated with said at least one image region is malignant and a likelihood that the part of the selected tissue associated with said at least one image region is benign.

5. A system as set forth in claim 1 wherein said processor determines a plurality of features for each of said image regions, each of said plurality of features having a value indicative of the characteristic of an associated one of said image regions, said processor applying said classifier to said plurality of features for at said at least some of said image regions to provide a likelihood value for said at least some of said image regions, each of the likelihood values indicating a likelihood that the part of the selected tissue associated with each of said at least some of said image regions has the known tissue condition.

6. A system as set forth in claim 5 wherein said processor is programmed to quantify the characteristic of each of said at least some of said image regions by fitting each corresponding characteristic to a predetermined curve, each of said plurality of features for each of said at least some of said image regions having a value indicative of a parameter of the predetermined curve quantifying the corresponding characteristic.

7. A system as set forth in claim 1 wherein said image set is a plurality of thermal images of the selected tissue of the patient and the characteristic of each of said image regions is a temperature profile for the part of the selected tissue associated with each of the image regions over the time period.

8. A system as set forth in claim 7 wherein at least a substantial number of said plurality of thermal images are acquired during application of a thermal challenge to the selected tissue of the patient, the characteristic of each of said image regions varying based on the temperature condition of the part of the selected tissue associated with said image region in response to the thermal challenge.

9. A system as set forth in claim 8 wherein the thermal challenge is a cooling challenge, the characteristic of each of said image regions representing a cooling response of the part of the selected tissue associated with a respective said image region.

10. A system as set forth in claim 9 wherein said processor is programmed to quantify the characteristic of each of said image regions by fitting the characteristic of each of said image regions to a predetermined model, said processor determining at least one feature value for each of said image regions indicative of a parameter of the corresponding predetermined model.

11. A system as set forth in claim 10 wherein at least part of the predetermined model for each of said image regions varies exponentially as a function of temperature of the associated part of the selected tissue over the time period.

12. A system as set forth in claim 11 wherein at least part of the predetermined model is:

13. A system for helping to determine a condition of selected tissue of a patient, said system comprising:

memory storing a plurality of thermal images of the selected tissue of the patient obtained over an image acquisition time period, the selected tissue having been exposed to a cooling challenge during a substantial part of the image acquisition time period, each of said thermal images including a plurality of image regions corresponding to an associated part of the selected tissue, each of said image regions having a characteristic that is variable over the image acquisition time period indicative of a temperature condition of the associated part of the selected tissue; and

a processor connected with said memory, said processor being operative to determine at least one feature for at least one of said image regions over the image acquisition time period, said at least one feature having a value indicative of the characteristic of said at least one image region over the image acquisition time period, said processor providing an indication of the condition of the part of the selected tissue associated with said at least one image region upon applying a statistical classifier to the value of said at least one feature, said statistical classifier describing a statistical distribution of a plurality of predetermined feature values derived from thermal images of previously diagnosed tissue having a known tissue condition.

14. A method for helping determine a condition of selected tissue of a patient, said method comprising the steps of:

acquiring an image set of the selected tissue over an image acquisition time period, the image set including a plurality of image regions, each image region having a characteristic that is variable over the image acquisition time period and indicative of a temperature condition of an associated part of the selected tissue;

applying a thermal challenge to the selected tissue during a substantial part of the image acquisition time period;

modeling the characteristic of at least one image region to quantify a plurality of features for each characteristic being modeled, each feature having a value functionally related to the characteristic being modeled;

selecting at least one feature for the at least one image region; and

applying a statistical classifier to each of the selected at least one feature to provide an indication of the condition of the part of the selected tissue associated with the at least one image region, the statistical classifier describing a plurality of predetermined feature values derived from image sets of previously diagnosed tissue having a predetermined tissue condition.

15. A method as set forth in claim 14 wherein said thermal challenge is a cooling challenge.

16. A method as set forth in claim 14 wherein said step of modeling further includes fitting at least part of the characteristic of the at least one image region to a predetermined model defined by an equation that varies exponentially as a function of temperature for the associated part of the selected tissue over the time period, the features for each characteristic being selected from the predetermined model.

17. A method as set forth in claim 16 wherein at least part of the predetermined model is:

18. A method as set forth in claim 14 wherein each of the image regions is at least one image pixel.

19. A method as set forth in claim 14 wherein said step of applying each of the selected at least one feature to the classifier further includes determining a likelihood that the part of the selected tissue associated with the at least one image region has the known tissue condition.

20. A method as set forth in claim 19 wherein a first part of the classifier describes a distribution of predetermined feature values derived from thermal image sets of previously diagnosed tissue having a first tissue condition, a second part of the classifier describes another distribution of predetermined feature values derived from thermal image sets of previously diagnosed tissue having a second tissue condition which is different from the first tissue condition, said method further including determining (i) a likelihood that the part of the selected region of tissue associated with the at least one image region has the first tissue condition and (ii) a likelihood that the part of the selected region of tissue associated with the at least one image region has the second tissue condition.

21. A method as set forth in claim 20 wherein the first tissue condition is malignant and the second tissue condition is benign.

22. A method of developing image data for use in helping determine a condition of selected tissue of a patient, said method comprising the steps of:

providing thermal image data for a plurality of image sets, each of the plurality of image sets being obtained over an image acquisition time period for a different selected tissue sample;

selecting an image region for each of the plurality of image sets corresponding to part of each selected tissue sample having a predetermined tissue condition, each selected image region having a characteristic that is variable over the image acquisition time period indicative of a temperature condition for the part of the selected tissue sample of the corresponding image set;

determining a plurality of features for each selected image region, each feature having a value indicative of the characteristic of a respective one of the selected image regions;

selecting at least one feature for each selected image region having a common predetermined tissue condition; and

developing a statistical classifier that describes a statistical distribution based on the values of each selected at least one feature, the statistical distribution being indicative of temperature characteristic for tissue having the common predetermined tissue condition.

23. A method as set forth in claim 22 wherein said step of determining further includes modeling the characteristic for each selected image region over the image acquisition time period to fit a predetermined model, each modeled characteristic defining at least some of the plurality of features for each selected image region based on corresponding parameters of the predetermined model.

24. A method as set forth in claim 23 wherein the predetermined model is:

25. A method as set forth in claim 23 wherein a thermal challenge is applied to each selected tissue sample during a substantial part of the image acquisition time period for each of the plurality of image sets, the features for each selected image region indicating changes in the temperature condition of the part of the selected tissue sample associated with each selected image region in response to the thermal challenge.

26. A method as set forth in claim 25 wherein the statistical distribution is a first statistical distribution for a class having a first predetermined tissue condition, said method further including the steps of:

selecting a second image region for each of another plurality of image sets corresponding to part of each selected tissue sample having a second predetermined tissue condition, each selected second image region having a characteristic that is variable over the image acquisition time period indicative of a temperature condition for the part of the selected tissue sample of the corresponding image set;

determining a plurality of features for each selected second image region, each feature having a value indicative of a respective one of the selected second image regions; and

developing the classifier based on the first statistical distribution and a second statistical distribution, the second statistical distribution being based on at least one feature selected for each of the selected second image regions having the second predetermined tissue condition.

27. A method as set forth in claim 26 wherein the first predetermined tissue condition is malignant and the second predetermined tissue condition is benign.

28. A method as set forth in claim 26 wherein each of the plurality of image sets includes a plurality of image frames, each selected image region further including a plurality of image subregions in each image frame, each of the plurality of image subregions in each image set having a value indicative of a thermal condition for an associated spatial region of the selected tissue sample that is variable over the image acquisition time period.

29. A method as set forth in claim 28 wherein the thermal characteristic of each of the plurality of image subregions is modeled to fit an equation that varies exponentially as a function of temperature for the associated spatial region of each respective the selected tissue sample over at least part of the corresponding image acquisition time period.

30. A method as set forth in claim 28 further including determining an average value of the thermal characteristic for each of the image subregions of a selected image region in each image frame and modeling a temporal response for each selected region of each image set using the determined average values, each temporal response indicating an average temperature condition for an associated spatial region of the respective selected tissue sample in each frame of the corresponding image acquisition time period, the plurality of features for each selected image region being determined based on the temporal response modeled for each respective selected image region.

31. A method as set forth in claim 30 wherein each subregion is at least one pixel.

32. A system for helping determine a condition of selected tissue of a patient, said system comprising:

means for obtaining an image set of the selected tissue over an image acquisition time period, the image set including a plurality of image regions, each image region having a characteristic that is variable over the image acquisition time period and indicative of a temperature condition of an associated part of the selected tissue;

means for applying a thermal challenge to the selected tissue during a substantial part of the image acquisition time period;

means for modeling the characteristic of at least one image region to quantify a plurality of features for each characteristic being modeled, each feature having a value functionally related to the characteristic being modeled;

means for selecting at least one feature for the at least one image region; and

means for applying a statistical classifier to each of the selected at least one feature to provide an indication of the condition of the part of the selected tissue associated with the at least one image region, the statistical classifier describing a plurality of predetermined feature values derived from image sets of previously diagnosed tissue having a predetermined tissue condition.