Systems and methods for segmenting object of interest from medical image
View Patent ↗A system for segmenting a target organ tumor from an image includes a background model builder, a foreground model builder and a tumor region locator. The background model builder uses an intensity distribution estimate of voxels in an organ region in an image to build a background model. The foreground model builder uses an intensity distribution estimate of voxels in a target organ tumor to build a first foreground model. The tumor region locator uses the background model and the first foreground model to segment the target organ tumor to obtain a first segmentation result.
1. A system for segmenting a target organ tumor from an image, comprising:
a background model builder, wherein the background model builder uses an intensity distribution estimate of voxels in an organ region in an image to build a background model;
a foreground model builder, wherein the foreground model builder uses an intensity distribution estimate of voxels in a target organ tumor to build a first foreground model; and
a tumor region locator, wherein the tumor region locator uses the background model and the first foreground model to segment the target organ tumor to obtain a first segmentation result.
2. The system of claim 1 , wherein the background model builder receives user input marking one or more sample organ regions in the image, and wherein the background model builder uses the sample organ regions for estimating the intensity distribution of voxels in the organ region in the image.
3. The system of claim 1 , wherein the background model builder marks organ regions in the image based on the estimated intensity distribution of voxels in the organ region and based on at least one of morphological operations, connected component analysis or convex hull analysis.
4. The system of claim 1 , wherein the foreground model builder receives user input marking a target organ tumor in the image, and wherein the foreground model builder uses the user input for estimating the intensity distribution of voxels in the target organ tumor.
5. The system of claim 1 , wherein the foreground model builder uses a Bayesian rule to build a likelihood field of the target organ tumor using the background model and the estimated intensity distribution of voxels in the target organ tumor.
6. The system of claim 1 , wherein the tumor region locator starts from a user-input position in the image to locate substantially all voxels in the target organ tumor.
7. The system of claim 1 , wherein the tumor region locator segments the target organ tumor using a likelihood field of the target organ tumor.
8. The system of claim 1 , wherein the foreground model builder uses the first segmentation result to build a second foreground model.
9. The system of claim 8 , wherein the tumor region locator uses the background model and the second foreground model to segment the target organ tumor to obtain a second segmentation result.
10. The system of claim 1 , wherein the organ region is one of a liver region, a pancreas region, a kidney region or a lung region.
11. A computer readable medium embodying instructions executed by a processor to perform a method of segmenting an object of interest from a medical image, the method comprising:
building a background using a segmentation technique;
building a foreground using the background and a block of sample voxel intensities of an object of interest in an image; and
segmenting the object of interest in an organ region of the image using the foreground and the background.
12. The computer readable medium of claim 11 , wherein building the background comprises receiving user input marking one or more samples from an organ region in an image.
13. The computer readable medium of claim 11 , wherein the segmentation technique is an organ segmentation technique.
14. The computer readable medium of claim 13 , wherein the organ segmentation technique comprises classifying voxels as organ or non-organ using an estimate of an intensity distribution of voxels in an organ region of the image.
15. The computer readable medium of claim 14 , wherein at least one of morphological operations, connected component analysis, or convex hull analysis is used to reduce classification errors.
16. The computer readable medium of claim 11 , the method further comprising building an estimate of an intensity distribution of voxels in the organ region using a statistical technique.
17. The computer readable medium of claim 16 , wherein the statistical technique is one of a nonparametric Parzen window technique, a histogram, a density estimation technique, or a Gaussian model.
18. The computer readable medium of claim 11 , wherein the block of sample voxel intensities comprises a small block of sample voxel intensities.
19. A system for providing segmentation of an object of interest from a medical image, comprising:
a memory device for storing a program;
a processor in communication with the memory device, the processor operative with the program to:
build a background using a segmentation technique;
build a foreground using the background and a block of sample voxel intensities of an object of interest in an image; and
segment the object of interest in an organ region of the image using the foreground and the background.
20. The system of claim 19 , wherein when building the background, the processor is further operative with the program to receive user input marking one or more samples from an organ region in an image.
21. The system of claim 19 , wherein when segmenting the object of interest the processor is further operative with the program to classify voxels as organ or non-organ using an estimate of an intensity distribution of voxels in an organ region of the image.