Segmentation of structures based on curvature slope

• US 20050196041A1
• Filed: 03/02/2005
• Published: 09/08/2005
• Est. Priority Date: 03/04/2004
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Bogoni, Luca, Lakare, Sarang, Jerebko, Anna
• Granted Patent: US 7,634,133 B2
• Time in Patent Office: Days
• US Class Current: 382/173
• CPC Class Codes: G06T 7/10 Segmentation; Edge detection motion-based segmentation G06T7/215
• CPC Class Codes: G06V 10/443 by matching or filtering

Abstract

A system and method for segmentation of structures in an image based on curvature slope are provided, the system including a processor, a curvature slope unit in signal communication with the processor for computing curvature slopes in the image and discarding voxels having a slope not satisfying specifications, and a segmentation unit in signal communication with the processor for locating connected clusters of retained voxels; and the method including computing curvature slopes in the image, discarding voxels having a slope not satisfying specifications, and locating connected clusters of retained voxels.

Claims

1. A method for segmentation of structures in an image based on curvature slope, the method comprising:

computing curvature slopes in the image;

discarding voxels having a slope not satisfying specifications; and

locating connected clusters of retained voxels.

2. A method as defined in claim 1, further comprising:

detecting edges within a volume of interest (VOI);

computing curvatures for surface voxels corresponding to the edges;

computing at least one of a minimum and maximum curvature slope for each cluster; and

selecting the cluster having the lowest minimum or highest maximum curvature slope for the segmentation structure.

3. A method as defined in claim 2 wherein the curvature slopes are computed at the surface voxels.

4. A method as defined in claim 1 wherein the specifications for discarding voxels specify those voxels with slopes outside of a desired range.

5. A method as defined in claim 1, further comprising:

determining the underlying surface for the connected clusters of retained voxels by interpolation; and

selecting all voxels between the original surface voxels and the underlying surface to be included in the segmented structure.

6. A method as defined in claim 1 wherein computing curvatures includes computing at least one of Gaussian and Mean curvatures.

7. A method as defined in claim 1 wherein computing curvature slopes includes:

computing average curvatures within concentric rings;

finding the ring with minimum or maximum average curvature; and

computing the curvature slope from the central ring to one of the peripheral rings having minimum or maximum average curvature.

8. A method as defined in claim 4 wherein the desired range of slopes for discarding voxels is bounded by zero.

9. A method as defined in claim 2 wherein selecting the cluster includes selecting the cluster having at least one of the lowest minimum Gaussian slope and the highest maximum Mean slope.

15. A system as defined in claim 9 wherein the curvature slope unit is disposed for discarding voxels outside of predefined specifications.

16. A system as defined in claim 15 wherein the predefined specifications include a desired range of slopes that is bounded by zero.

10. A system for segmentation of structures in an image based on curvature slope, the system comprising:

a processor;

a curvature slope unit in signal communication with the processor for computing curvature slopes in the image and discarding voxels having a slope not satisfying specifications; and

a segmentation unit in signal communication with the processor for locating connected clusters of retained voxels.

11. A system as defined in claim 10, further comprising:

interpolation means in signal communication with the segmentation unit for determining the underlying surface for the connected clusters of retained voxels by interpolation; and

selection means in signal communication with the interpolation means for selecting all voxels between an original surface and the underlying surface to be included in the segmented structure.

12. A system as defined in claim 10, further comprising at least one of an imaging adapter and a communications adapter in signal communication with the processor for receiving image data.

13. A system as defined in claim 10 wherein the curvature slope unit is disposed for computing curvatures as average curvatures for all surface voxels in at least one concentric ring relative to a central surface voxel.

14. A system as defined in claim 10 wherein the curvature slope unit is disposed for computing curvatures as at least one of Gaussian and Mean curvatures.

17. A system as defined in claim 10 wherein the curvature slope unit is disposed for computing at least one of a minimum and maximum curvature slope for each cluster by computing at least one of a minimum Gaussian slope and a maximum Mean slope for each cluster.

18. A system as defined in claim 11 wherein the selection means is disposed for selecting the cluster having at least one of the lowest minimum Gaussian slope and the highest maximum Mean slope.

19. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform program steps for segmentation of structures in an image based on curvature slope, the program steps comprising:

computing curvature slopes in the image;

discarding voxels having a slope not satisfying specifications; and

locating connected clusters of retained voxels.

20. A device as defined in claim 19, the program steps further comprising at least one of:

detecting edges within a volume of interest (VOI);

computing curvatures for surface voxels corresponding to the edges;

computing curvature slopes at the surface voxels;

discarding voxels having a slope outside of specifications;

locating connected clusters of retained voxels;

computing at least one of a minimum and maximum curvature slope for each cluster;

selecting the cluster having at least one of the lowest minimum and highest maximum curvature slope for the segmentation structure;

determining the underlying surface for the segmentation structure by interpolation;

selecting all voxels between the original surface voxels and the underlying surface to be included in the segmented structure

computing a minimum Gaussian slope for each cluster; and

computing a maximum Mean slope for each cluster.