System and methods for automated segmentation of individual skeletal bones in 3D anatomical images

Patent metadata

Field	Value
Publication number	US-10178982-B2
Application number	US-201815985070-A
Country	US
Kind code	B2
Filing date	May 21, 2018
Priority date	Jul 29, 2015
Publication date	Jan 15, 2019
Grant date	Jan 15, 2019

Abstract

Presented herein, in certain embodiments, are approaches for robust bone splitting and segmentation in the context of small animal imaging, for example, microCT imaging. In certain embodiments, a method for calculating and applying single and hybrid second-derivative splitting filters to gray-scale images and binary bone masks is described. These filters can accurately identify the split lines/planes of the bones even for low-resolution data, and hence accurately morphologically disconnect the individual bones. The split bones can then be used as seeds in region growing techniques such as marker-controlled watershed segmentation. With this approach, the bones can be segmented with much higher robustness and accuracy compared to prior art methods.

First claim

What is claimed is:

1. A method of performing image segmentation to automatically differentiate individual organs in an image of a subject, the method comprising:
receiving, by a processor, a gray-scale image of a subject;
applying, by the processor, a plurality of second derivative splitting filters to the image to produce split organ masks for the image based on each second derivative splitting filter;
combining the split organ masks for the image produced by each second derivative splitting filter into a hybrid split organ mask for the image, using one or more logical operations;
determining, by the processor, a plurality of split binary components of the hybrid split organ mask by performing one or more morphological processing operations; and
performing, by the processor, a region growing operation using the split binary components of the hybrid split organ mask as seeds, thereby producing a segmentation map differentiating individual organs in the image.

2. The method of claim 1 , wherein the plurality of second derivative splitting filters comprises members selected from the group comprising of a Laplacian of Gaussian (LoG), a highest Hessian eigenvalue with preliminary Gaussian filtering (HEH), and a lowest Hessian eigenvalue with preliminary Gaussian filtering (HEH).

3. The method of claim 1 , wherein, for each of the plurality of second derivative splitting filters being applied, producing a filtered image and identifying voxels of the filtered image with intensity higher or lower than a predetermined threshold, thereby yielding a split binary mask identifying voxels in the vicinity of a boundary region between adjacent organs; and
wherein combining the split organ masks using one or more logical operations generates a hybrid split organ mask with voxels near organ boundaries removed.

4. The method of claim 1 , further comprising:
identifying, by the processor, one or more anatomical measurements using the segmentation map.

5. The method of claim 1 , further comprising:
receiving feedback from a user identifying one or more segmented regions of the segmentation map for further refinement or recalculation and then:
applying, by the processor, one or more second derivative splitting filters to portions of the image corresponding to the one or more user-identified segmented regions to produce one or more additional split organ masks;
determining, by the processor, split binary components for each of the additional split organ masks; and
performing, by the processor, a region growing operation using the split binary components of the additional split organ masks as seeds, thereby producing a refined or recalculated segmentation map differentiating individual organs in the image.

6. The method of claim 1 , wherein determining the plurality of split binary components of the hybrid split organ mask further comprises performing connected component analysis and/or identifying catchment basins using distance and watershed transforms.

7. The method of claim 1 , wherein at least one of the plurality of second derivative splitting filters is applied using one or more rotational invariants of spatial second partial derivatives of the voxel intensities of the image.

8. The method of claim 1 , wherein the image is an in vivo image.

9. The method of claim 1 , wherein the subject is a small mammal subject.

10. The method of claim 1 , wherein the greyscale image is received from a micro-CT imager.

11. The method of claim 1 , wherein the organ comprises one or more bones.

12. A system comprising:
a processor; and
a memory having instructions stored thereon, wherein the instructions, when executed by the processor, cause the processor to:
receive a gray-scale image of a subject;
apply a plurality of second derivative splitting filters to the image to produce split organ masks for the image based on each second derivative splitting filter;
combine the split organ masks for the image produced by each second derivative splitting filter into a hybrid split organ mask for the image, using one or more logical operations;
determine a plurality of split binary components of the hybrid split organ mask by performing one or more morphological processing operations; and
perform a region growing operation using the split binary components of the hybrid split organ mask as seeds, thereby producing a segmentation map differentiating individual organs in the image.

13. The system of claim 12 , wherein the plurality of second derivative splitting filters comprises members selected from the group comprising of a Laplacian of Gaussian (LoG), a highest Hessian eigenvalue with preliminary Gaussian filtering (HEH), and a lowest Hessian eigenvalue with preliminary Gaussian filtering (HEH).

14. The system of claim 12 ,
wherein, for each of the plurality of second derivative splitting filters being applied, producing a filtered image and identifying voxels of the filtered image with intensity higher or lower than a predetermined threshold, thereby yielding a split binary mask identifying voxels in the vicinity of a boundary region between adjacent organs; and
wherein combining the split organ masks using one or more logical operations generates a hybrid split organ mask with voxels near organ boundaries removed.

15. The system of claim 12 , wherein the instructions, when executed by the processor, further cause the processor to identify one or more anatomical measurements using the segmentation map.

16. The system of claim 12 , wherein the instructions, when executed by the processor, further cause the processor to,
receive feedback from a user identifying one or more segmented regions of the segmentation map for further-refinement or recalculation and then:
apply one or more second derivative splitting filters to portions of the image corresponding to the one or more user-identified segmented regions to produce one or more additional split organ masks;
determine split binary components for each of the additional split organ masks; and
perform a region growing operation using the split binary components of the additional split organ masks as seeds, thereby producing a refined or recalculated segmentation map differentiating individual organs in the image.

17. The system of claim 12 , wherein the determining the plurality of split binary components of the hybrid split organ mask further comprises performing connected component analysis and/or identifying catchment basins using distance and watershed transforms.

18. The system of claim 12 , wherein at least one of the plurality of second derivative splitting filters is applied using one or more rotational invariants of spatial second partial derivatives of the voxel intensities of the image.

19. A non-transitory-computer readable medium having instructions stored thereon, wherein the instructions, when executed by a processor, cause the processor to:
receive a gray-scale image of a subject;
apply a plurality of second derivative splitting filters to the image to produce split organ masks for the image based on each second derivative splitting filter;
combine the split organ masks for the image produced by each second derivative splitting filter into a hybrid split organ mask for the image, using one or more logical operations;
determine a plurality of split binary components of the hybrid split organ mask by performing one or more morphological processing operations; and
perform a region growing operation using the split binary components of the hybrid split organ mask as seeds, thereby p

Assignees

• Perkinelmer Health Sci Inc
• Perkinelmer Cellular Tech Germany Gmbh

Inventors

• Behrooz Ali
• Kask Peet

Classifications

• G06T2207/30008

  Bone · CPC title

• G06T15/00

  Three-dimensional [3D] image rendering · CPC title

• A61B6/505 Primary

  for diagnosis of bone · CPC title

• G06K2009/4666

  Physics · mapped topic

• A61B6/508

  for non-human patients · CPC title