System and method for image segmentation

Patent metadata

Field	Value
Publication number	US-10282844-B2
Application number	US-201615320466-A
Country	US
Kind code	B2
Filing date	May 5, 2016
Priority date	May 5, 2015
Publication date	May 7, 2019
Grant date	May 7, 2019

Abstract

A system and method for image segmentation are provided. A three-dimensional image data set representative of a region including at least one airway may be acquired. The data set may include a plurality of voxels. A first-level seed within the region may be identified. A first-level airway within the region may be identified based on the first-level seed. A second-level airway may be identified within the region based on the first-level airway. The first-level airway and the second-level airway may be fused to form an airway tree.

First claim

What is claimed is:

1. A method of image processing implemented on at least one machine, each of which has at least one processor and storage device, the method comprising:
acquiring a three-dimensional image data set representative of a region including at least one airway, the data set comprising a plurality of voxels;
identifying a first-level seed within the region;
identifying a first-level airway within the region based on the first-level seed;
defining a plurality of second-level seeds based on the first-level airway;
identifying a second-level airway within the region by performing growing based on the plurality of second-level seeds using an energy-based 3D reconstruction algorithm; and
fusing the first-level airway and the second-level airway to form an airway tree.

2. The method of claim 1 , the identifying a first-level airway comprising:
selecting, from the plurality of voxels, an adjacent voxel of the first-level seed and neighboring voxels of the adjacent voxel;
determining a characteristic value for the adjacent voxel and a characteristic value for each of the neighboring voxels;
calculating a difference between the characteristic value for the adjacent voxel and the characteristic value for each of the neighboring voxels; and
marking, when the characteristic value of the adjacent voxel is below a first threshold and when, for each of the neighboring voxels, the difference is below a second threshold, the adjacent voxel as part of the first-level airway.

3. The method of claim 1 , the identifying a first-level airway comprising:
determining a voxel set based on a level set method;
selecting a voxel from the voxel set;
determining a first characteristic value of the voxel, wherein the first characteristic value includes a grey level; and
marking, when the first characteristic value of the voxel is below a third threshold, the voxel as part of the first-level airway.

4. The method of claim 1 , the performing growing comprising:
selecting a plurality of adjacent voxels for each of the plurality of second-level seeds;
initializing a voxel set;
for each adjacent voxel,
calculating a first growing potential energy of the adjacent voxel at a first state;
calculating a second growing potential energy of the adjacent voxel at a second state;
determining, based on the first growing potential energy and the second growing potential energy, whether to add the adjacent voxel into the voxel set;
determining whether the voxel set belongs to the second-level airway; and
adding, when the voxel set is determined to belong to the second-level airway, the voxel set into the second-level airway.

5. The method of claim 4 , the performing growing further comprising:
defining, when the voxel set is determined to belong to the second-level airway, a new seed among the voxel set.

6. The method of claim 4 , the determining whether the voxel set belongs to the second-level airway comprising:
identifying a connected domain within the voxel set;
determining a dimension of the connected domain;
determining a voxel count of the connected domain;
performing a first comparison between the dimension of the connected domain and a fifth threshold;
performing a second comparison between the voxel count of the connected domain and a sixth threshold; and
determining, based on the first comparison or the second comparison, whether the connected domain belongs to the second-level airway.

7. The method of claim 6 further comprising:
removing, when the connected domain is determined not to belong to the second-level airway, the connected domain from the voxel set.

8. The method of claim 6 further comprising:
determining, based on the first comparison and the second comparison, a voxel not belonging to the second-level airway; and
removing, when the voxel is determined not to belong to the second-level airway, the voxel from the voxel set.

9. The method of claim 4 , wherein the first growing potential energy is associated with at least one first weighting factor, the second growing potential energy is associated with at least one second weighting factor, and the method further includes:
adjusting, when the voxel set is determined not to belong to the second-level airway, the at least one first weighting factor, or the at least one second weighting factor, or both.

10. The method of claim 1 , wherein the first-level airway includes a low-level bronchus of a lung region, and the second-level airway includes a terminal bronchiole of a lung region.

11. The method of claim 1 further comprising:
identifying a lung organ;
identifying a lung tissue based on the fused airway and the lung organ; and
labelling a left lung tissue or a right lung tissue of the identified lung tissue based on the fused airway.

12. The method of claim 1 further comprising:
identifying a lung organ;
obtaining an image for a 2D layer of the lung organ;
smoothing the image based on a morphology based method; and
labelling a left lung portion of the lung organ including the second-level airway or a right lung portion of the lung organ including the second-level airway based on the smoothed image and the first-level airway.

13. The method of claim 11 , the identifying a lung organ comprising:
determining a lung region based on the plurality of voxels;
removing a background based on the determined lung region;
determining a 2D layer based on area of the 2D layer; and
performing 3D region growing based on the 2D layer.

14. The method of claim 13 , the determining a lung region comprising:
selecting, from the plurality of voxels, a voxel; and
determining, based on a comparison between a characteristic value of the voxel and a seventh threshold, whether the voxel belongs to the lung region.

15. The method of claim 13 , the removing a background comprising:
identifying a background seed near an edge of the lung region; and
performing region growing based on the background seed.

16. A non-transitory computer readable medium comprising executable instructions that, when executed by at least one processor, cause the at least one processor to effectuate a method comprising
acquiring a three-dimensional image data set representative of a region including at least one airway, the data set comprising a plurality of voxels;
identifying a first-level seed within the region;
identifying a first-level airway within the region based on the first-level seed;
defining a plurality of second-level seeds based on the first-level airway;
identifying a second-level airway within the region by performing growing based on the plurality of second-level seeds using an energy-based 3D reconstruction algorithm; and
fusing the first-level airway and the second-level airway to form an airway tree.

17. A system comprising:
at least one processor, and
executable instructions that, when executed by the at least one processor, cause the at least one processor to effectuate a method comprising
acquiring a three-dimensional image data set representative of a region including at least one airway, the data set comprising a plurality of voxels;
identifying a first-level seed within the region;
identifying a first-level airway within the region based on the first-level seed;
defining a plurality of second-level seeds based on the first-level airway;
identifying a second-level airway within the region by performing growing based on the plurality of second-level seeds using an energy-based 3D reconstruction algorithm; and
fusing the first-level airway and the second-level ai

Assignees

• Shanghai United Imaging Healthcare Co Ltd

Inventors

• Wang Lilong
• Wang Lijia
• Mao Yufei
• Meng Xiaolin
• Wang Xiaodong
• Li Cheng

Classifications

• G06T7/11 Primary

  Region-based segmentation · CPC title

• G06T2207/10072

  Tomographic images · CPC title

• G06T2200/04

  involving 3D image data · CPC title

• G06T7/187

  involving region growing; involving region merging; involving connected component labelling · CPC title

• G06T7/155

  involving morphological operators · CPC title