Methods and systems for image segmentation

Patent metadata

Field	Value
Publication number	US-10949977-B2
Application number	US-201816236954-A
Country	US
Kind code	B2
Filing date	Dec 31, 2018
Priority date	Jun 30, 2017
Publication date	Mar 16, 2021
Grant date	Mar 16, 2021

Abstract

The application discloses a method and system for segmenting a lung image. The method may include obtaining a target image relating to a lung region. The target image may include a plurality of image slices. The method may also include segmenting the lung region from the target image, identifying an airway structure relating to the lung region, and identifying one or more fissures in the lung region. The method may further include determining one or more pulmonary lobes in the lung region.

First claim

What is claimed is:

1. A method implemented on at least one machine which has at least one processor and at least one storage device, the method comprising:
obtaining a target image relating to a lung region, the target image including a plurality of image slices;
segmenting the lung region from the target image;
identifying an airway structure relating to the lung region, comprising:
identifying one or more fissures in the lung region; and
determining, based on the airway structure and the one or more fissures, one or more pulmonary lobes in the lung region, wherein the identifying one or more fissures in the lung region comprises:
identifying a plurality of candidate fissures in the lung region;
sorting at least a part of the plurality of candidate fissures based on a sorting algorithm;
determining a plurality of values, each of the plurality of values corresponding to one candidate fissure of the plurality of candidate fissures, wherein the plurality of values includes a plurality of normal directions corresponding to the plurality of candidate fissures;
classifying, based on the plurality of values, the plurality of candidate fissures into at least two groups; and
merging candidate fissures belonging to a same group.

2. The method of claim 1 , wherein the segmenting the lung region from the target image comprises:
identifying a characteristic slice among the plurality of image slices;
determining a starting slice and an end slice based on the characteristic slice;
determining the lung region based on the starting slice and the end slice; and
determining a left lung or a right lung in the lung region.

3. The method of claim 2 , wherein the identifying a characteristic slice among the plurality of image slices comprises:
determining a distribution of CT values of the plurality of image slices; and
selecting, based on the distribution, the characteristic slice, wherein the CT values vary in a range.

4. The method of claim 1 , wherein the identifying an airway structure relating to the lung region comprises:
obtaining a reference point of the airway structure;
generating, based on the reference point, an airway tree corresponding to the airway structure; and
identifying one or more branches of the airway tree, wherein a branch of the one or more branches of the airway tree corresponds to a branch of the airway structure.

5. The method of claim 4 , wherein the identifying one or more branches of the airway tree further comprises:
identifying at least one node of the airway tree, the node relating to a trachea, a left main bronchus, a right main bronchus, a left upper lobe bronchus, a left lower lobe bronchus, a right upper lobe bronchus or a right lower lobe bronchus.

6. The method of claim 1 , wherein the identifying a plurality of candidate fissures in the lung region comprises:
determining a Hessian matrix corresponding to a voxel of the lung region;
determining an Eigenvalue of the Hessian matrix; and
designating, based on the Eigenvalue, that the voxel belongs to a candidate fissure of the plurality of candidate fissures.

7. The method of claim 1 , wherein the identifying one or more fissures in the lung region further comprises fitting the merged candidate fissures to obtain the one or more fissures in the lung region.

8. The method of claim 1 , wherein the determining one or more pulmonary lobes comprises:
determining a distance from a point of the lung region to the airway structure; and
determining, based on the distance, a pulmonary lobe to which the point of the lung region belongs, wherein the point of the lung region is located in an expansion region of the one or more fissures.

9. A system for image segmentation, comprising:
a lung segmentation module configured to obtain a target image relating to a lung region and segment the lung region in the target image;
an airway identification module configured to identify an airway structure relating to the lung region;
a fissure segmentation module configured to identify one or more fissures in the lung region; and
a pulmonary lobe segmentation module configured to determine one or more pulmonary lobes based on the airway structure and the one or more fissures, wherein to identify one or more fissures in the lung region, the system is directed to:
identify a plurality of candidate fissures in the lung region;
sort at least a part of the plurality of candidate fissures based on a sorting algorithm;
determine a plurality of values, each of the plurality of values corresponding to one candidate fissure of the plurality of candidate fissures, wherein the plurality of values includes a plurality of normal directions corresponding to the plurality of candidate fissures;
classify the plurality of candidate fissures into at least two groups based on the plurality of values; and
merge candidate fissures belonging to a same group.

10. The system of claim 9 , wherein the target image includes a plurality of image slices, and the lung segmentation module is configured to:
identify a characteristic slice among the plurality of image slices;
determine a starting slice and an end slice based on the characteristic slice;
determine the lung region based on the starting slice and the end slice; and
determine a left lung or a right lung in the lung region.

11. The system of claim 10 , wherein to identify a characteristic slice among the plurality of image slices, the system is directed to:
determine a distribution of CT values of the plurality of image slices; and
select the characteristic slice based on the distribution, wherein the CT values vary in a certain range.

12. The system of claim 9 , wherein the airway identification module is configured to:
obtain a reference point of the airway structure;
generate an airway tree corresponding to the airway structure based on the reference point; and
identify one or more branches of the airway tree, wherein a branch of the one or more branches of the airway tree corresponds to a branch of the airway structure.

13. The system of claim 9 , wherein to identify a plurality of candidate fissures in the lung region, the system is directed to:
determine a Hessian matrix corresponding to a voxel of the lung region;
determine an Eigenvalue of the Hessian matrix; and
designate that the voxel belongs to a candidate fissure of the plurality of candidate fissures based on the Eigenvalue.

14. The system of claim 9 , wherein to identify one or more fissures in the lung region, the system is further directed to fit the merged candidate fissures to obtain the one or more fissures in the lung region.

15. The system of claim 9 , wherein the pulmonary lobe segmentation module is configured to:
determine a distance from a point of the lung region to the airway structure; and
determine, based on the distance, a pulmonary lobe to which the point of the lung region belongs, wherein the point of the lung region is located in an expansion region of the one or more fissures.

16. A non-transitory computer readable medium, storing instructions, the instructions when executed by a processor, causing the processor to execute operations comprising:
obtaining a target image relating to a lung region, the target image including a plurality of image slices;
segmenting the lung region from the target image;
identifying an airway structure relating to the lung region;
identifying one or more fissures in the lung region; and
determining, based on the airway structure and the one or more fissures, one or more pulmonary lobes in the lung region, wherein the identify

Assignees

• Shanghai United Imaging Healthcare Co Ltd

Inventors

• Wang Xiaodong
• Mao Yufei
• Jin Renchao
• Tian Yujie
• Hu Naiwen
• Xu Lijun
• He Fengli
• Liu Hong
• He Kai
• Song Enmin
• Xu Xiangyang

Classifications

• G06F18/24

  Classification techniques · CPC title

• G06T2207/20156

  Automatic seed setting · CPC title

• G06T7/136

  involving thresholding · CPC title

• G06T7/12

  Edge-based segmentation · CPC title

• G06T2207/30172

  Centreline of tubular or elongated structure · CPC title