Multi-phase registration of 2-D X-ray images to 3-D volume studies

• US 20070127845A1
• Filed: 11/16/2005
• Published: 06/07/2007
• Est. Priority Date: 11/16/2005
• Status: Active Grant

• Current Assignee: Accuray Incorporated
• Original Assignee: Accuray Incorporated
• Inventors: Kuduvalli, Gopinath, Fu, Dongshan
• Granted Patent: US 7,835,500 B2
• Time in Patent Office: Days
• US Class Current: 382/294
• CPC Class Codes: A61B 6/4458 the source unit or the detector unit being attached to robotic arms
• CPC Class Codes: A61B 6/4464 the source unit or the detector unit being mounted to ceiling
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 7/32 using correlation-based methods
• CPC Class Codes: G06T 7/38 Registration of image sequences
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: Y10S 128/922 including image analysis

Abstract

A method for registering 2-D radiographic images to 3-D volume studies includes generating digitally reconstructed radiographs (DRRs) from a 3-D volume study. The DRR'"'"'s are used as reference images to register a patient'"'"'s position during image-guided radiosurgery. Stereoscopic X-rays of the patient are acquired during treatment, and multi-phase registration is used to register the DRRs with the X-ray images in two or more projections.

Claims

1. A method, comprising:

estimating transformation parameters between a 2-D projection of a volume and a 2-D reference image of the volume in a plurality of search-phases; and

searching for 2-D transformation parameters in each of the plurality of search-phases using combinations of search techniques and similarity measures in each search phase.

2. The method of claim 1, further comprising repeating a search for one or more transformation parameters to achieve a desired registration accuracy between the 2-D projection of the volume and the 2-D reference image of the volume.

3. The method of claim 1, further comprising converting the transformation parameters into a 3-D rigid transformation of the volume.

4. The method of claim 2, wherein estimating the transformation parameters between the 2-D projection of the volume and the 2-D reference image of the volume comprises:

estimating in-plane transformation parameters between the 2-D projection of the volume and a 2-D reference image generated from 3-D scan data of the volume, using a first similarity measure and a first search method in each of two or more projections; and

estimating out-of-plane rotation parameters between the 2-D projection and a plurality of 2-D reference images generated from the 3-D scan data of the volume, using a second similarity measure and a second search method in each of the two or more projections.

5. The method of claim 4, wherein repeating a search for one or more transformation parameters to achieve a desired registration accuracy between the 2-D projection of the volume and the 2-D reference image of the volume comprises refining the in-plane transformation parameters and the out-of-plane rotation parameters using the second similarity measure and at least a third search method in each of the two or more projections.

6. The method of claim 4, wherein the first similarity measure comprises a simple similarity measure and the first search method comprises a hierarchical search.

7. The method of claim 6, wherein the hierarchical search comprises 3-parameter multi-level matching.

8. The method of claim 6, wherein the first similarity measure is one of a sum of squared differences (SSD) measure and a sum of absolute differences (SAD) measure.

9. The method of claim 5, wherein the out-of-plane rotation parameters in each of the two or more projections comprise two out-of-plane rotation parameters, and wherein the second search method comprises a 1-D search in each of the two or more projections.

10. The method of claim 9, wherein the 1-D search searches the two out-of-plane rotation parameters in each of the two or more projections.

11. The method of claim 9, wherein one of the out-of-plane rotation parameters in each projection comprises one of the in-plane rotation parameters in another projection, and wherein the 1-D search searches only one of the two out-of-plane rotation parameters in each of the two or more projections.

12. The method of claim 9, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein refining the in-plane transformation parameters in each projection comprises:

refining the in-plane translation parameters in each projection using 2-D sub-pixel matching; and

refining the in-plane rotation parameter in each projection using the 1-D search.

13. The method of claim 12, further comprising:

determining whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy; and

repeating the method of claim 10 if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy;

else

refining the out-of-plane rotation parameters in each projection using the 1-D search.

14. The method of claim 9, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein refining the in-plane transformation parameters and the out-of-plane rotation parameters comprises:

refining the in-plane translation parameters in each projection using a steepest descent minimization;

refining the in-plane rotation parameters in each projection using the steepest descent minimization; and

refining the out-of-plane rotation parameters in each projection using the 1-D search.

15. The method of claim 14, further comprising:

determining whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy; and

repeating the method of claim 14 if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy;

else

refining the out-of-plane rotation parameters in each projection using the 1-D search.

16. An article of manufacture, including machine-accessible instructions that when accessed by a data processing system, cause the data processing system to perform a method, comprising:

estimating transformation parameters between a 2-D projection of a volume and a 2-D reference image of the volume in a plurality of search-phases; and

searching for 2-D transformation parameters in each of the plurality of search-phases using combinations of search techniques and similarity measures in each search phase.

17. The article of manufacture of claim 16, further comprising repeating a search for one or more transformation parameters to achieve a desired registration accuracy between the 2-D projection of the volume and the 2-D reference image of the volume.

18. The method of claim 16, further comprising converting the transformation parameters into a 3-D rigid transformation of the volume.

19. The method of claim 17, wherein estimating the transformation parameters between the 2-D projection of the volume and the 2-D reference image of the volume comprises:

estimating in-plane transformation parameters between the 2-D projection of the volume and a 2-D reference image generated from 3-D scan data of the volume, using a first similarity measure and a first search method in each of two or more projections; and

estimating out-of-plane rotation parameters between the 2-D projection and a plurality of 2-D reference images generated from the 3-D scan data of the volume, using a second similarity measure and a second search method in each of the two or more projections.

20. The article of manufacture of claim 19, the method further comprising refining the in-plane transformation parameters and the out-of-plane rotation parameters using the second similarity measure and at least a third search method in each of the two or more projections.

21. The article of manufacture of claim 19, wherein the first similarity measure comprises a SSD similarity measure and the first search method comprises a hierarchical search.

22. The article of manufacture of claim 21, wherein the hierarchical search comprises 3-parameter multi-level matching.

23. The article of manufacture of claim 21, wherein the first similarity measure is one of a sum of squared differences (SSD) measure and a sum of absolute differences (SAD) measure.

24. The article of manufacture of claim 20, wherein the out-of-plane rotation parameters in each of the two or more projections comprise two out-of-plane rotation parameters, and wherein the second search method comprises a 1-D search in each of the two or more projections.

25. The article of manufacture of claim 24, wherein the 1-D search searches the two out-of-plane rotation parameters in each of the two or more projections.

26. The article of manufacture of claim 24, wherein one of the out-of-plane rotation parameters in each projection comprises one of the in-plane rotation parameters in another projection, and wherein the 1-D search searches only one of the two out-of-plane rotation parameters in each of the two or more projections.

27. The article of manufacture of claim 24, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein refining the in-plane transformation parameters in each projection comprises:

refining the in-plane translation parameters in each projection using 2-D sub-pixel matching and; and

refining the in-plane rotation parameter in each projection using the 1-D search.

28. The article of manufacture of claim 27, the method further comprising:

determining whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy; and

repeating the method of claim 24 if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy;

else

refining the out-of-plane rotation parameters in each projection using the 1-D search.

29. The article of manufacture of claim 24, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein refining the in-plane transformation parameters and the out-of-plane rotation parameters comprises:

refining the in-plane translation parameters in each projection using a steepest descent minimization;

refining the in-plane rotation parameters in each projection using the steepest descent minimization; and

refining the out-of-plane rotation parameters in each projection using the 1-D search.

30. The article of manufacture of claim 29, the method further comprising:

determining whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy; and

repeating the method of claim 26 if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy;

else

refining the out-of-plane rotation parameters in each projection using the 1-D search.

31. A system comprising a processing device, wherein the processing device is configured to estimate transformation parameters between a 2-D projection of a volume and a 2-D reference image of the volume in a plurality of search-phases, and wherein the processing device is further configured to search for 2-D transformation parameters in each of the plurality of search-phases using combinations of search techniques and similarity measures in each search phase.

32. The system of claim 31, wherein the processing device is further configured to repeat a search for one or more transformation parameters to achieve a desired registration accuracy between the 2-D projection of the volume and the 2-D reference image of the volume.

33. The system of claim 31, wherein the processing device is further configured to convert the transformation parameters into a 3-D rigid transformation of the volume.

34. The system of claim 32, wherein to estimate the transformation parameters between the 2-D projection of the volume and the 2-D reference image of the volume, the processing device is configured to estimate in-plane transformation parameters between the 2-D projection of the volume and the 2-D reference image generated from the 3-D scan data of the volume, using a first similarity measure and a first search method in each of two or more projections, and wherein the processing device is further configured to estimate out-of-plane rotation parameters between the 2-D projection and the plurality of 2-D reference images using a second similarity measure and a second search method in each of the two or more projections.

35. The system of claim 34, further comprising a data storage device coupled with the data processing device to store a plurality of 2-D projections of the volume and a plurality of 2-D reference images generated from 3-D scan data of the volume.

36. The system of claim 35, wherein the processing device is further configures to refine the in-plane transformation parameters and the out-of-plane rotation parameters using a second similarity measure and at least a third search method in each of the two or more projections.

37. The system of claim 35, wherein the first similarity measure comprises a SSD similarity measure and the first search method comprises a hierarchical search.

38. The system of claim 37, wherein the hierarchical search comprises 3-parameter multi-level matching.

39. The system of claim 37, wherein the first similarity measure is one of a sum of squared differences (SSD) measure and a sum of absolute differences (SAD) measure.

40. The system of claim 36, wherein the out-of-plane rotation parameters in each of the two or more projections comprise two out-of-plane rotation parameters, and wherein the second search method comprises a 1-D search in each of the two or more projections.

41. The system of claim 44, wherein the 1-D search searches the two out-of-plane rotation parameters in each of the two or more projections.

42. The system of claim 40, wherein one of the out-of-plane rotation parameters in each projection comprises one of the in-plane rotation parameters in another projection, and wherein the 1-D search searches only one of the two out-of-plane rotation parameters in each of the two or more projections.

43. The system of claim 35, wherein the second similarity measure comprises an optimized pattern intensity similarity measure.

44. The system of claim 40, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein to refine the in-plane transformation parameters in each projection, the processing device is further configured to refine the in-plane translation parameters in each projection using 2-D sub-pixel matching and to refine the in-plane rotation parameter in each projection using the 1-D search.

45. The system of claim 44, wherein the processing device is further configured to determine whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy, to further refine the in-plane transformation parameters in each projection if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy, else to refine the out-of-plane rotation parameters in each projection using the 1-D search.

46. The system of claim 40, wherein the in-plane transformation parameters comprise two in-plane translations and one in-plane rotation in each projection, wherein the second similarity measure comprises an optimized pattern intensity similarity measure, and wherein to refine the in-plane transformation parameters and the out-of-plane rotation parameters the processing device is further configured to refine the in-plane translation parameters in each projection using a steepest descent minimization, to refine the in-plane rotation parameters in each projection using the steepest descent minimization, and to refine the out-of-plane rotation parameters in each projection using the 1-D search.

47. The system of claim 46, wherein the processing device is further configured to determine whether the in-plane translation parameters and the in-plane rotation parameters in each projection have enough accuracy to achieve a desired registration accuracy, and to further to refine the in-plane transformation parameters and the out-of-plane rotation parameters if the in-plane translation parameters and the in-plane rotation parameters do not have enough accuracy to achieve the desired registration accuracy, else to refine the out-of-plane rotation parameters in each projection using the 1-D search.

48. The system of claim 35, further comprising:

an imaging system coupled with the data storage device to generate the plurality of 2-D projection of the volume; and

a treatment planning system to generate the plurality of 2-D reference images from the 3-D scan data of the volume, the plurality of 2-D reference images comprising a nominal in-plane reference image and a plurality of out-of-plane reference images in each of the two or more projections.

49. An apparatus, comprising:

means for estimating transformation parameters between a 2-D projection of a volume and a 2-D reference image of the volume in a plurality of search-phases; and

means for searching for 2-D transformation parameters in each of the plurality of search-phases using combinations of search techniques and similarity measures in each search phase.

50. The apparatus of claim 49, further comprising means for repeating a search for one or more transformation parameters to achieve a desired registration accuracy between the 2-D projection of the volume and the 2-D reference image of the volume.

51. The apparatus of claim 49, further comprising means for converting the transformation parameters into a 3-D rigid transformation of the volume.