Automatic quality checks for radiotherapy contouring

Patent metadata

Field	Value
Publication number	US-10290093-B2
Application number	US-201615273454-A
Country	US
Kind code	B2
Filing date	Sep 22, 2016
Priority date	Sep 22, 2015
Publication date	May 14, 2019
Grant date	May 14, 2019

Abstract

Systems, devices, methods, and computer processing products for automatically checking for errors in segmentation (contouring) using heuristic and/or statistical evaluation methods.

First claim

The invention claimed is:

1. A method for detecting radiation therapy segmentation errors in one or more segments of an anatomical structure of a patient, comprising:
generating one or more segments of the anatomical structure by delineating the anatomical structure on radiation therapy image slices obtained for the patient; and
evaluating the one or more segments using one or more evaluation methods to determine whether the one or more segments pass one or more predetermined evaluation criteria,
wherein the evaluating includes evaluating the one or more segments using a heuristic evaluation method,
wherein the evaluating using the heuristic evaluation method comprises:
evaluating the one or more segments using a set of heuristic rules; and
determining whether at least one of the heuristic rules is violated,

wherein, if at least one of the heuristic rules is violated, a determination is made that an error in the segmentation is present,
wherein the determining whether at least one of the heuristic rules is violated includes evaluating the one or more segments using Boolean operations and pixel/voxel counting.

2. The method of claim 1 , wherein the evaluating is automatic.

3. The method of claim 1 , further comprising:
generating binary images for the one or more segments;
combining the binary images using at least one of a plurality of Boolean operators, to thereby obtain a resulting binary image; and
determining whether the resulting binary image satisfies the heuristic rule by counting pixels/voxels in the resulting binary image.

4. The method of claim 3 , wherein a heuristic rule is violated if the number of pixels/voxels is not within a predetermined range.

5. The method of claim 4 , further comprising displaying a signal alerting that a segmentation error is present.

6. The method of claim 4 , further comprising displaying a segmentation report including the results of the segmentation evaluation.

7. The method of claim 1 , wherein the determining whether at least one of the heuristic rules is violated further includes evaluating the one or more segments using a 3D connectivity labeling process.

8. The method of claim 1 , wherein the evaluating further includes evaluating the one or more segments using a statistical evaluation method, which comprises evaluating a shape of a segment and determining whether the shape of the segment is within a predetermined range of known shapes for the segment.

9. The method of claim 8 , further comprising generating a shape model based on a probabilistic distribution of shape variations found in the range of known shapes for the segment, and performing a statistical test to check if the shape of the segment is within the predetermined range.

10. The method of claim 9 , wherein the probabilistic distribution includes a multivariate normal distribution (μ, Σ).

11. The method of claim 10 , wherein the performing of the statistical check comprises evaluating how well the segment shape fits within the multivariate normal distribution using a probabilistic principal component analysis (PPCA).

12. The method of claim 11 , wherein if the shape of the segment does not fit within the multivariate normal distribution within a predetermined value, it is determined that a segmentation error is present.

13. The method of claim 1 , wherein the evaluating is in real time.

14. A system for automatically detecting errors in a radiation therapy segmentation of an anatomical structure, comprising:
a radiation therapy imaging device configured to generate or obtain one or more image slices containing the anatomical structure, the radiation therapy imaging device being further configured to generate one or more segments on the image slices by delineating the anatomical structure on the image slices; and
a computer processing device configured to automatically evaluate the generated segments using one or more evaluation methods,
wherein the computer processing device comprises a database including a set of heuristic rules,
wherein the automatic evaluation includes checking whether one or more of the segments fail at least one of the heuristic rules,
wherein the computer processing device is further configured to generate binary images for the segments and to combine the binary images using one or more Boolean operators to obtain a combined binary image, and
wherein the computer processing device is further configured to evaluate whether the resulting binary image complies with a heuristic rule using a pixel/voxel counting process.

15. The system of claim 14 , further comprising displaying an error signal on a display device if the computer processing device determines that the resulting binary image fails to comply with a heuristic rule.

16. The system of claim 14 , wherein the automatic evaluation further includes evaluating the one or more segments using a statistical evaluation method, wherein the computer processing device is further configured to implement a statistical shape model to evaluate a shape of a segment and determine whether the shape of the segment is within a predetermined range of known shapes for the segment.

17. A method for detecting radiation therapy segmentation errors in one or more segments of an anatomical structure, comprising:
generating one or more segments of the anatomical structure by delineating the anatomical structure on radiation therapy image slices obtained by imaging a patient; and
evaluating the one or more segments using a statistical evaluation method including:
evaluating a shape of the segment; and
determining whether the shape of the segment is within a predetermined range of known shapes for the segment,

wherein the evaluating further includes:
generating a shape model based on a probabilistic distribution of shape variations found in the range of known shapes for the segment; and
performing a statistical test to check if the shape of the segment is within the predetermined range.

18. The method of claim 17 , wherein the probabilistic distribution includes a multivariate normal distribution (μ, Σ).

19. The method of claim 18 , wherein the performing of the statistical check comprises evaluating how well the segment shape fits within the multivariate normal distribution using a probabilistic principal component analysis (PPCA).

20. The method of claim 19 , wherein if the shape of the segment does not fit within the multivariate normal distribution within a predetermined value, it is determined that a segmentation error is present.

Assignees

• Varian Medical Systems Int Ag

Inventors

• Albrecht Thomas
• Hess Peter
• Houhou Nawal

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06K2209/05

  Physics · mapped topic

• G06T2207/10116

  X-ray image · CPC title

• G06T2207/20076

  Probabilistic image processing · CPC title

• G06T2207/30008

  Bone · CPC title