Method for segmenting medical images and detecting surface anomalies in anatomical structures

• US 6,556,696 B1
• Filed: 02/05/2002
• Issued: 04/29/2003
• Est. Priority Date: 08/19/1997
• Status: Expired due to Term

• Current Assignee: Health and Human Services United States of America As Represented By The SEC
• Original Assignee: United States Department Of Health And Human Services
• Inventors: Summers, Ronald M., Selbie, Scott, Malley, James D., Pusanik, Lynne M.
• Primary Examiner(s): Johns, Andrew W.
• Assistant Examiner(s): NAKHJAVAN, SHERVIN K
• Application Number:
  US10/072,667
• Time in Patent Office: 448 Days
• Field of Search: 382/128, 382/131, 382/154, 382/203, 382/204, 382/206, 382/218, 382/220, 382/225, 382/277, 382/285, 382/286, 345/420, 345/424, 345/425, 345/441, 345/442, 345/443, 434/403, 434/211, 434/256, 434/257, 700/29, 700/31, 700/98, 700/118
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/20101 Interactive definition of point of interest, landmark or seed
• CPC Class Codes: G06T 2207/30028 Colon; Small intestine
• CPC Class Codes: G06T 2207/30061 Lung
• CPC Class Codes: G06T 2207/30064 Lung nodule
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/187 involving region growing; involving region merging; involving connected component labelling
• CPC Class Codes: G06V 20/64 Three-dimensional objects

Abstract

A region growing method segments three-dimensional image data of an anatomical structure using a tortuous path length limit to constrain voxel growth. The path length limit constrains the number of successive generations of voxel growth from a seed point to prevent leakage of voxels outside the boundary of the anatomical structure. Once segmented, a process for detecting surface anomalies performs a curvature analysis on a computer model of the surface of the structure. This process detects surface anomalies automatically by traversing the vertices in the surface model, computing partial derivatives of the surface at the vertices, and computing curvature characteristics from the partial derivatives. To identify possible anomalies, the process compares the curvature characteristics with predetermined curvature characteristics of anomalies and classifies the vertices. The process further refines potential anomalies by segmenting neighboring vertices that are classified as being part of an anomaly using curvature characteristics. Finally, the process colorizes the anomalies and computes a camera position and direction for each one to assist the user in viewing 2D renderings of the computer model.

Claims

1. A computer-assisted method for detecting anomalies in a model representing a three-dimensional anatomical structure, the method comprising:

computing a curvature characteristic in a local neighborhood around a vertex in the model;

classifying curvature of a cluster of connected vertices by comparing the computed curvature characteristic to a predetermined curvature classification; and

defining an anomaly as a cluster of connected vertices having a predetermined curvature classification.

2. The method of claim 1 further comprising:

computing partial derivatives during traversal of the model representing the three-dimensional anatomical structure;

wherein the curvature characteristic is computed from the partial derivatives.

3. The method of claim 2, wherein

the model representing the anatomical structure comprises a plurality of vertices;

and

computing partial derivatives comprises computing partial derivatives at the vertices.

4. The method of claim 2 wherein computing partial derivatives comprises convolving a filter over discrete voxel data at neighboring voxels.

5. The method of claim 2 wherein the curvature characteristics comprise Gaussian curvatures computed from the partial derivatives.

6. The method of claim 2 wherein the curvature characteristics comprise curvatures chosen from one or more of the following and computed from the partial derivatives:

Gaussian curvatures;

mean curvatures; and

principal curvatures.

7. The method of claim 2 wherein computing partial derivatives comprises employing a 3D filtering technique.

8. The method of claim 7 wherein computing partial derivatives comprises employing a kernel in the 3D filtering technique.

9. The method of claim 1 further comprising:

before computing the curvature characteristic around the vertex in the model, fitting a parametric patch to the local neighborhood around the vertex in the model.

10. The method of claim 9 wherein the parametric patch comprises a b-spline patch.

11. The method of claim 1 wherein the local neighborhood comprises one or more isosurfaces.

12. The method of claim 1 wherein the local neighborhood comprises a wall of the anatomical structure.

13. The method of claim 1 wherein

the model comprises a three-dimensional surface model.

14. The method of claim 1 wherein

the three-dimensional anatomical structure comprises a colon;

and

the anomaly comprises a polyp.

15. A computer-readable medium comprising computer-executable instructions for performing the following to detect anomalies in a model representing a three-dimensional anatomical structure:

computing a curvature characteristic in a local neighborhood around a vertex in the model;

classifying curvature of a cluster of connected vertices by comparing the computed curvature characteristic to a predetermined curvature classification; and

defining an anomaly as a cluster of connected vertices having a predetermined curvature classification.

16. An apparatus operable to detect anomalies in a model representing a three-dimensional anatomical structure, the apparatus comprising:

means for computing a curvature characteristic in a local neighborhood around a vertex in the model;

means for classifying curvature of a cluster of connected vertices by comparing the computed curvature characteristic to a predetermined curvature classification; and

means for defining an anomaly as a cluster of connected vertices having a predetermined curvature classification.

17. A method of detecting anomalies in an anatomical structure, the method comprising:

computing a representation of the anatomical structure;

computing curvature characteristics for the representation of the anatomical structure;

comparing the curvature characteristics with predetermined characteristics associated with anomalies; and

identifying an anomaly in the representation of the anatomical structure based at least on the comparing.

18. The method of claim 17 further comprising:

computing partial derivatives during traversal of the representation of the anatomical structure;

wherein the curvature characteristics are computed from the partial derivatives.

19. The method of claim 18 wherein computing partial derivatives comprises employing a 3D filtering technique.

20. The method of claim 19 wherein computing partial derivatives comprises employing a kernel in the 3D filtering technique.

21. The method of claim 18 wherein

the representation of the anatomical structure comprises a plurality of vertices;

and

traversal of the representation of the anatomical structure comprises visiting the vertices.

22. The method of claim 18 wherein

the representation of the anatomical structure comprises a plurality of vertices;

and

computing partial derivatives comprises computing partial derivatives at the vertices.

23. The method of claim 18 wherein computing partial derivatives comprises convolving a filter over discrete voxel data at neighboring voxels.

24. The method of claim 18 wherein the curvature characteristics comprise Gaussian curvatures computed from the partial derivatives.

25. The method of claim 18 wherein the curvature characteristics comprise curvatures chosen from one or more of the following and computed from the partial derivatives:

Gaussian curvatures;

mean curvatures; and

principal curvatures.

26. The method of claim 17 wherein computing curvature characteristics comprises computing a curvature characteristic in a local neighborhood around a vertex in the representation.

27. The method of claim 26 further comprising:

before computing the curvature characteristic around the vertex, fitting a parametric patch to the local neighborhood around the vertex in the model.

28. The method of claim 17 wherein the representation of the anatomical structure comprises a plurality of vertices, the method further comprising:

classifying a cluster of vertices in the representation of the anatomical structure as an anomaly.

29. The method of claim 17 wherein the anomaly comprises a lesion.

30. The method of claim 17 wherein the anomaly comprises a polypoid lesion.

31. The method of claim 17 wherein curvature characteristics are computed from a representation of a surface of the anatomical structure.

32. The method of claim 17 wherein computing a representation of the anatomical structure comprises computing an isosurface of the anatomical structure.

33. The method of claim 17 wherein

the representation of the anatomical structure represents a wall of a bronchus captured from a scan of a human patient;

and

the anomaly represents a lesion detected on the wall.

34. The method of claim 17 wherein

the anatomical structure comprises a colon;

and

the anomaly represents a polyp detected in the colon.

35. The method of claim 17 further comprising:

displaying a depiction of the representation of the anatomical structure;

wherein identified anomalies are colored to distinguish from background anatomy.

36. The method of claim 17 further comprising:

presenting a guided virtual tour of the anatomical structure, repositioning a viewpoint to a plurality of identified anomalies.

37. The method of claim 17, wherein predetermined characteristics indicating elliptical curvature of peak subtype are associated with surface anomalies.

38. A computer-readable medium comprising computer-readable instructions for performing the following to detect anomalies in an anatomical structure:

computing a representation of the anatomical structure;

computing curvature characteristics for the representation of the anatomical structure;

comparing the curvature characteristics with predetermined characteristics associated with anomalies; and

identifying an anomaly in the representation of the anatomical structure based at least on the comparing.

39. A computer-implemented method comprising:

computing a representation of a bronchus comprising a plurality of vertices;

computing partial derivatives during traversal of the vertices of the representation of the bronchus via a 3D filter;

computing curvature characteristics from the partial derivatives;

based on the curvature characteristics, classifying a shape; and

identifying a polypoid lesion in the representation of the bronchus based on having classified the shape as elliptical with a peak subtype.

40. An apparatus operable to detect anomalies in an anatomical structure, the apparatus comprising:

means for computing a representation of the anatomical structure;

means for computing curvature characteristics for the representation of the anatomical structure;

means for comparing the curvature characteristics with predetermined characteristics associated with anomalies; and

means for identifying an anomaly in the representation of the anatomical structure based at least on the comparing.