Semi-automatic segmentation algorithm for pet oncology images

• US 7,006,677 B2
• Filed: 04/15/2002
• Issued: 02/28/2006
• Est. Priority Date: 04/15/2002
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Manjeshwar, Ravindra Mohan, Cline, Harvey Ellis
• Primary Examiner(s): Werner, Brian
• Assistant Examiner(s): LAVIN, CHRISTOPHER L
• Application Number:
  US10/122,892
• Publication Number:
  US 20030194119A1
• Time in Patent Office: 1,415 Days
• Field of Search: 382/128, 382/164, 382/165, 382/170, 382/171, 382/173, 382/180, 382/181, 382/199, 382/224, 382/225
• US Class Current: 382/131
• CPC Class Codes: G06T 2207/10104 Positron emission tomography [PET]
• CPC Class Codes: G06T 2207/20104 Interactive definition of region of interest [ROI]
• CPC Class Codes: G06T 2207/20156 Automatic seed setting
• CPC Class Codes: G06T 2207/30016 Brain
• 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 10/248 by interactive preprocessing or interactive shape modelling, e.g. feature points assigned by a user
• CPC Class Codes: G06V 10/267 by performing operations on regions, e.g. growing, shrinking or watersheds
• CPC Class Codes: Y10S 128/922 including image analysis

Abstract

An apparatus and method for segmenting three-dimensional (3D) medical images containing a region of interest is provided that identifies a first set of seed points within the region of interest and a second set of seed points outside the region of interest. A first sphere is constructed within the region of interest. Voxels contained within the medical image are classified using a spatial constrained fuzzy clustering algorithm. A plurality of second spheres is generated. Ones of the plurality of second spheres are accepted that satisfy the homogeneity function threshold as defined by the spatial constricted fuzzy clustering algorithm. A three-dimensional area is grown that defines the region of interest. The region of interest defined by the three-dimensional area is displayed.

Claims

1. A method for segmenting three-dimensional (3D) medical images containing a region of interest, the method comprising the steps of:

identifying a first set of seed points within the region of interest;

identifying a second set of seed points outside the region of interest;

constructing a first sphere within the region of interest centered around the first set of seed points;

classifying voxels contained within 3D medical images using a spatial constrained fuzzy clustering algorithm whereby transforming the voxels contained within the 3D medical image into a fuzzy partition domain based on a homogeneity function;

successively generating a plurality of second spheres about said first sphere;

accepting ones of the plurality of second spheres that satisfy the homogeneity function threshold as defined by the spatial constricted fuzzy clustering algorithm;

adaptively growing a three-dimensional area defining the region of interest based on the step of accepting; and

displaying the region of interest defined by the step of adaptively growing.

2. The method of claim 1 wherein the three-dimensional images are acquired by at least one of magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), and x-ray systems.

3. The method of claim 1 further comprising the step of selecting a radius of curvature of the first sphere and the plurality of second spheres to eliminate noise voxels on the interface between an inside of the region of interest and outside the region of interest.

4. The method of claim 1 wherein the first spheres and the plurality of second spheres have a selected radius based on a predetermined radius of curvature.

5. A method for segmenting three-dimensional (3D) medical images containing a region of interest, the method comprising the steps of:

identifying a first set of seed points within the region of interest;

identifying a second set of seed points outside the region of interest;

constructing a first sphere within the region of interest centered around the first set of seed points.

classifying voxels contained within 3D medical images using a spatial constrained fuzzy clustering algorithm whereby transforming the voxels contained within the 3D medical image into a fuzzy partition domain based on a homogeneity function;

successively generating a plurality of second spheres about said first sphere;

accepting ones of the plurality of second spheres that satisfy the homogeneity function threshold as defined by the spatial constrained fuzzy clustering algorithm;

adaptively growing a three-dimensional area defining the region of interest based on the step of accepting;

selecting a radius of curvature of the first sphere and the plurality of second spheres to eliminate noise voxels on the interface between an inside of the region of interest and outside the region of interest; and

displaying the region of interest defined by the steps of adaptively growing and selecting the radius of curvature.

6. The method of claim 5 wherein the three-dimensional (3D) images are acquired by at least one of magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), and x-ray systems.

7. The method of claim 5 wherein the first sphere and the plurality of second spheres have a selected radius based on a predetermined radius of curvature.

8. A system for segmenting a medical image containing a region of interest and acquired by an image acquisition device, the system comprising:

a processor coupled to the image acquisition device, the processor computing segmentation of the medical image wherein the computation of segmentation comprises the steps of;

classifying voxels contained within the medical image using a spatial constrained fuzzy clustering algorithm as having a homogeneity function whereby transforming the voxels contained within the medical image to a fuzzy partition domain;

constructing a first sphere within the region of interest wherein the first sphere is centered about a first set of seed points;

successively generating a plurality of second spheres about said first sphere;

accepting ones of the plurality of second spheres that satisfy the homogeneity function threshold as defined by the spatial constrained fuzzy clustering algorithm;

adaptively growing a three-dimensional area defining the region of interest based on the step of accepting; and

displaying the region of interest defined by the steps of adaptively growing; and

an interface unit coupled to the processor for interpreting information relating to the segmentation of the medical image.

9. The system of claim 8 wherein the processor determines a radius of curvature of the first sphere and the plurality of second spheres to eliminate noise voxels on the interface between an inside of the region of interest and outside the region of interest.

10. The system of claim 8 wherein the interface unit receives first and second seed point data from a user of the system.

11. The system of claim 8 wherein the first sphere and the plurality of second spheres have a selected radius based on a predetermined radius of curvature.

12. The system of claim 8 wherein the image acquisition device is at least one of magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), and x-ray systems.

13. A system for segmenting medical images containing a region of interest and acquired by an image acquisition device, the system comprising:

a processor coupled to the image acquisition device, the processor computing segmentation of the medical image wherein the computation of segmentation comprises the steps of;

constructing a first sphere within the region of interest wherein the first sphere is centered about a first set of seed points;

classifying voxels contained within the medical image using a spatial constrained fuzzy clustering algorithm as having a homogeneity function whereby transforming the voxels contained within the medical image to a fuzzy partition domain;

successively generating a plurality of second spheres about said first sphere;

accepting ones of the plurality of second spheres that satisfy the homogeneity function threshold as defined by the spatial constrained fuzzy clustering algorithm;

adaptively growing a three-dimensional area defining the region of interest based on the step of accepting;

determining a radius of curvature of the first sphere and the plurality of second spheres to eliminate noise voxels on the interface between an inside of the region of interest and outside the region of interest; and

displaying the region of interest defined by the steps of adaptively growing and selecting the radius of curvature;

an interface unit coupled to the processor for interpreting information relating to the segmentation of the medical image.

14. The system of claim 13 wherein the interface unit receives first and second seed point data from a user of the system.

15. The system of claim 13 wherein the first sphere and the plurality of second spheres have a selected radius based on a predetermined radius of curvature.

16. The system of claim 13 wherein the image acquisition device is at least one of magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), and x-ray systems.