Method and system for semantics driven image registration

1. A method for automatic semantic driven registration of medical images, comprising:

detecting anatomic landmarks and organs in a first image and a second image;

automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images by mapping the semantic information found in the at least one text-based document to a predetermined ontology; and

registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information.

2. The method of claim 1 , further comprising:

detecting pathologies in the first image and the second image.

3. The method of claim 2 , wherein the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

registering the second image to the first image based at least in part on the detected anatomic landmarks and organs, the detected pathologies, and the extracted semantic information.

4. The method of claim 1 , wherein the step of automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images comprises:

automatically searching the at least one text-based document for semantic information; and

mapping semantic information found in the at least one text-based document to a predetermined ontology.

5. The method of claim 1 , wherein the at least one text-based document comprises at least one of Radiology Information System (RIS) data, clinical reports, and DICOM header information.

6. The method of claim 1 , wherein the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest.

7. The method of claim 6 , wherein the step of enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest comprises:

increasing a weight of a similarity measure for pixels within the identified region of diagnostic interest.

8. The method of claim 1 , wherein the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

enforcing a greater precision on the registration of corresponding landmark pairs that are identified as structures of interest based on the extracted semantic information.

9. The method of claim 1 , wherein the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

automatically selecting a transformation model for transforming the second image to the first image based on the identified region of diagnostic interest.

10. The method of claim 1 , further comprising:

displaying the registered first and second images; and

refining the registration of the first and second images based on user interactions with the registered first and second images.

11. The method of claim 10 , wherein the step of refining the registration of the first and second images based on user interactions with the registered first and second images comprises:

shifting a focus of the registration of the first and second images based on a user selection of windowing settings.

12. The method of claim 10 , wherein the step of refining the registration of the first and second images based on user interactions with the registered first and second images comprises:

shifting a focus of the registration of the first and second images based on a user performing at least one of labeling or measurements in a particular image area.

13. The method of claim 10 , wherein the step of refining the registration of the first and second images based on user interactions with the registered first and second images comprises:

shifting a focus of the registration of the first and second images based on a user selection of a hyperlink in a displayed parsed clinical report.

14. The method of claim 10 , wherein the step of refining the registration of the first and second images based on user interactions with the registered first and second images comprises:

tracking eyes of a user viewing the registered first and second images;

and refining the registration based on the amount of time the eyes of the user have spent in different image areas.

15. An apparatus for automatic semantic driven registration of medical images, comprising:

means for detecting anatomic landmarks and organs in a first image and a second image;

means for automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images by mapping the semantic information found in the at least one text-based document to a predetermined ontology; and

means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information.

16. The apparatus of claim 15 , further comprising:

means for detecting pathologies in the first image and the second image.

17. The apparatus of claim 16 , wherein the means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs, the detected pathologies, and the extracted semantic information.

18. The apparatus of claim 15 , wherein the means for automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images comprises:

means for automatically searching the at least one text-based document for semantic information; and

means for mapping semantic information found in the at least one text-based document to a predetermined ontology.

19. The apparatus of claim 15 , wherein the means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

means for identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

means for enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest.

20. The apparatus of claim 19 , wherein the means for enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest comprises:

means for increasing a weight of a similarity measure for pixels within the identified region of diagnostic interest.

21. The apparatus of claim 15 , wherein the means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

means for enforcing a greater precision on the registration of corresponding landmark pairs that are identified as structures of interest based on the extracted semantic information.

22. The apparatus of claim 15 , wherein the means for registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprises:

means for identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

means for automatically selecting a transformation model for transforming the second image to the first image based on the identified region of diagnostic interest.

23. The apparatus of claim 15 , further comprising:

means for displaying the registered first and second images; and

means for refining the registration of the first and second images based on user interactions with the registered first and second images.

24. A non-transitory computer readable medium encoded with computer executable instructions for automatic semantic driven registration of medical images, the computer executable instructions defining steps comprising:

detecting anatomic landmarks and organs in a first image and a second image;

automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images by mapping the semantic information found in the at least one text-based document to a predetermined ontology; and

registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information.

25. The computer readable medium of claim 24 , further comprising computer executable instructions defining the step of:

detecting pathologies in the first image and the second image.

26. The computer readable medium of claim 25 , wherein the computer executable instructions defining the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprise computer executable instructions defining the step of:

registering the second image to the first image based at least in part on the detected anatomic landmarks and organs, the detected pathologies, and the extracted semantic information.

27. The computer readable medium of claim 24 , wherein the computer executable instructions defining the step of automatically extracting semantic information from at least one text-based document associated with at least one of the first and second images comprise computer executable instructions defining the steps of:

automatically searching the at least one text-based document for semantic information; and

mapping semantic information found in the at least one text-based document to a predetermined ontology.

28. The computer readable medium of claim 24 , wherein the computer executable instructions defining the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprise computer executable instructions defining the steps of:

identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest.

29. The computer readable medium of claim 28 , wherein the computer executable instructions defining the step of enforcing a greater precision on the registration of pixels of the first and second images in the identified region of diagnostic interest than pixels not in the identified region of diagnostic interest comprise computer executable instructions defining the step of:

increasing a weight of a similarity measure for pixels within the identified region of diagnostic interest.

30. The computer readable medium of claim 24 , wherein the computer executable instructions defining the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprise computer executable instructions defining the step of:

enforcing a greater precision on the registration of corresponding landmark pairs that are identified as structures of interest based on the extracted semantic information.

31. The computer readable medium of claim 24 , wherein the computer executable instructions defining the step of registering the second image to the first image based at least in part on the detected anatomic landmarks and organs and the extracted semantic information comprise computer executable instructions defining the step of:

identifying a region of diagnostic interest in the first and second images based on the extracted semantic information; and

automatically selecting a transformation model for transforming the second image to the first image based on the identified region of diagnostic interest.

32. The computer readable medium of claim 24 , further comprising computer executable instructions defining the steps of:

displaying the registered first and second images; and

refining the registration of the first and second images based on user interactions with the registered first and second images.