System and method for validating an image segmentation algorithm

• US 7,907,756 B2
• Filed: 01/09/2006
• Issued: 03/15/2011
• Est. Priority Date: 01/31/2005
• Status: Active Grant

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Zhang, Li, Zhang, Tiantian
• Primary Examiner(s): Alavi; Amir
• Application Number:
  US11/328,353
• Publication Number:
  US 20070081724A1
• Time in Patent Office: 1,891 Days
• Field of Search: 382128-134, 382/162, 382/164, 382/171, 382/173, 382/178, 382/187, 382218-220, 358/523, 358/525, 345/589, 345600-604
• US Class Current: 382/128
• CPC Class Codes: G06V 10/26 Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion

Abstract

A system and method for validating an image segmentation algorithm are provided. The method for validating an image segmentation algorithm comprises: determining a region of interest in an image; segmenting the image from a first point in the region of interest by using a computer-based segmentation algorithm to obtain a first segmentation result; segmenting the image from a second point in the region of interest by using the computer-based segmentation algorithm to obtain a second segmentation result; and comparing the first segmentation result with the second segmentation result to determine a consistency of the computer-based segmentation algorithm.

Claims

1. A method for guiding an initialization of an image segmentation, comprising:

identifying a region of interest (ROI) in an image, wherein the ROI includes a plurality of points;

segmenting the image using every point inside the ROI as an initial point to produce a segmentation result for each point in the ROI;

comparing the segmentation result of each point in the ROI to the segmentation result of every other point in the ROI to produce a consistency measure for each point in the ROI;

assigning a degree of color to each point in the ROI based on its consistency measure, wherein the degree of color indicates whether or not the point is a good candidate for an initial point of an image segmentation; and

displaying the image with different degrees of color in the ROI to provide a user with information as to where a segmentation of the image should be initialized.

2. The method of claim 1, wherein the ROI is identified manually or automatically.

3. The method of claim 1, wherein the image is acquired by using one of a CT, MR, positron emission tomography (PET), fluoroscopic, ultrasound, x-ray or hybrid imaging technique, or scanning or digital image recording technique.

4. The method of claim 1, wherein a first initial point of the points in the ROI is selected manually or automatically.

5. The method of claim 4, wherein a second initial point of the points in the ROI is selected manually or automatically.

6. The method of claim 1, wherein the consistency measure for a point in the ROI is determined by using a plurality of overlap ratios, which indicate an amount of overlap between the segmentation result of the point and the segmentation result of every other point in the ROI.

7. The method of claim 6, wherein a degree of color for a point in the ROI is computed from the point'"'"'s overlap ratios.

8. The method of claim 1, wherein when the image is a two-dimensional image each point in the ROI is a pixel and when the image is a three-dimensional image each point in the ROI is a voxel.

9. A system for guiding an initialization of an image segmentation, comprising:

a memory device for storing a program;

a processor in communication with the memory device, the processor operative with the program to;

indentify a region of interest ROI in an image, wherein the ROI includes a plurality of points;

segment the image using every point inside the ROI as an initial point to produce a segmentation result for each point in the ROI;

compare the segmentation result of each point in the ROI to the segmentation result of every other point in the ROI to produce a consistency measure for each point in the ROI;

assign a degree of color to each point in the ROI based on its consistency measure, wherein the degree of color indicates whether or not the point is a good candidate for an initial point of an image segmentation; and

display the image with different degrees of color in the ROI to provide a user with information as to where a segmentation of the image should he initialized.

10. The system of claim 9, wherein the ROI is identified manually or automatically.

11. The system of claim 9 wherein the image is acquired by using one of a CT, MR, positron emission tomography (PET), fluoroscopic, ultrasound, x-ray or hybrid imaging device, or digital camera or flatbed scanner.

12. The system of claim 9 wherein a first initial point of the points in the ROI is selected manually or automatically.

13. The system of claim 12 wherein the a second initial point of the points in the ROI is selected manually or automatically.

14. The system of claim 9, wherein the consistency measure for a point in the ROI is determined by using a plurality of overlap ratios, which indicate an amount of overlap between the segmentation result of the point and the segmentation result of every other point in the ROI.

15. The system of claim 14 wherein a degree of color for a point in the ROI is computed from the point'"'"'s overlap ratios.

16. A computer readable medium embodying a program of instructions executable by machine to perform steps for guiding an initialization of an image segmentation the steps comprising:

identifying a region of interest (ROI) in an image, wherein the ROI includes a plurality of points;

segmenting the image using every point inside the ROI as an initial point to produce a segmentation result for each point in the ROI;

comparing the segmentation result of each point in the ROI to the segmentation result of every other point in the ROI to produce a consistency measure for each point in the ROI;

assigning a degree of color to each point in the ROI based on its consistency measure, wherein the degree of color indicates whether or not the point is a good candidate for an initial point of an image segmentation; and

displaying the image with different degrees of color in the ROI to provide a user with information as to where a segmentation of the image should be initialized.

18. The computer readable medium method of claim 16, wherein the image is acquired by using one of a CT, MR, positron emission tomography (PET), fluoroscopic, ultrasound, x-ray or hybrid imaging technique, or scanning or digital image recording technique.

19. The computer readable medium of claim 16, wherein a first initial point of the points in the ROI is selected manually or automatically.

20. The computer readable medium of claim 16, wherein a second initial point of the points in the ROI is selected manually or automatically.

21. The computer readable medium of claim 16 wherein the consistency measure for a point in the ROI is determined by using a plurality of overlap ratios, which indicate an amount of overlap between the segmentation result of the point and the segmentation result of every other point in the ROI.

22. The computer readable medium of claim 21, wherein a degree of color for a point in the ROI is computed from the point'"'"'s overlap ratios.

17. The computer readable medium 16 wherein the ROI is identified manually or automatically.