System and method for auto-contouring in adaptive radiotherapy

1. A method, implemented by a processor, for performing auto-contouring in a medical image, the method comprising:

receiving the medical image, wherein the medical image contains a region of interest and a plurality of voxels;

identifying a first contour, by the processor, using a boundary detector, wherein the first contour is formed by a first set of voxels of the medical image located on a surface of the region of interest;

refining the first contour, by the processor, based on a shape dictionary, to generate a second contour of the region of interest; and

updating the boundary detector, by the processor, based on a second set of voxels located along the second contour.

2. The method of claim 1 , further including refining the second contour using the updated boundary detector.

3. The method of claim 1 , further comprising:

updating the shape dictionary based on the second set of voxels located along the second contour.

4. The method of claim 3 , further comprising:

refining the second contour using the updated shape dictionary.

5. The method of claim 3 , wherein updating the shape dictionary includes:

selecting a subset of shapes from the medical image containing the second contour;

obtaining a sparse coefficient for each shape in the subset; and

updating the shape dictionary based on the obtained sparse coefficients.

6. The method of claim 1 , wherein refining the first contour based on the shape dictionary to generate the second contour includes:

selecting a set of shapes from the shape dictionary;

combining the selected set of shapes to approximate the first contour; and

generating the second contour based on the combined set of shapes.

7. The method of claim 1 , further comprising training the boundary detector based on sample voxels selected from previously contoured medical images, wherein the sample voxels are located on known contours of the region of interest in the previously contoured medical images.

8. The method of claim 7 , wherein the boundary detector is trained using a Random Forest model.

9. The method of claim 7 , wherein updating the boundary detector includes re-training the boundary detector based on the second set of voxels.

10. A system for performing auto-contouring in a medical image, comprising:

a processor; and

a memory operatively coupled to the processor and storing instructions, that when executed by the processor, causes the processor to perform a method, the method comprising:

receiving the medical image, wherein the medical image contains a region of interest and a plurality of voxels;

identifying a first contour using a boundary detector, wherein the first contour is formed by a first set of voxels of the medical image located on a surface of the region of interest;

refining the first contour based on a shape dictionary to generate a second contour of the region of interest; and

updating the boundary detector based on a second set of voxels located along the second contour.

11. The system of claim 10 , wherein the method further includes refining the second contour using the updated boundary detector.

12. The system of claim 10 , wherein the method further comprises updating the shape dictionary based on the second set of voxels located along the second contour.

13. The system of claim 12 , wherein the method further comprises refining the second contour using the updated shape dictionary.

14. The system of claim 10 , wherein refining the first contour based on the shape dictionary to generate the second contour includes:

selecting a set of shapes from the shape dictionary;

combining the selected set of shapes to approximate the first contour; and

generating the second contour based on the combined set of shapes.

15. The system of claim 10 , wherein the method further comprises training the boundary detector based on sample voxels selected from previously contoured medical images, wherein the sample voxels are located on known contours of the region of interest in the previously contoured medical images.

16. A non-transitory computer-readable medium for storing instructions, that when executed by a processor, causes the processor to perform a method for performing auto-contouring in a medical image, the method comprising:

receiving the medical image, wherein the medical image contains a region of interest and a plurality of voxels;

identifying a first contour using a boundary detector, wherein the first contour is formed by a first set of voxels of the medical image located on a surface of the region of interest;

refining the first contour based on a shape dictionary to generate a second contour of the region of interest; and

updating the boundary detector based on a second set of voxels located along the second contour.

17. The non-transitory computer-readable medium of claim 16 , wherein the method further comprises training the boundary detector based on sample voxels selected from previously contoured medical images, wherein the sample voxels are located on known contours of the region of interest in the previously contoured medical images.

18. The non-transitory computer-readable medium of claim 17 , wherein updating the boundary detector includes re-training the boundary detector based on the second set of voxels.

19. The non-transitory computer-readable medium of claim 16 , wherein the method further comprises:

updating the shape dictionary based on the second set of voxels located along the second contour; and

refining the second contour using the updated shape dictionary.

20. The non-transitory computer-readable medium of claim 16 , wherein the method further includes refining the second contour using the updated boundary detector.