Method and device for generating anatomical labels for a physiological tree structure

Patent metadata

Field	Value
Publication number	US-10548552-B2
Application number	US-201816116889-A
Country	US
Kind code	B2
Filing date	Aug 29, 2018
Priority date	Dec 21, 2017
Publication date	Feb 4, 2020
Grant date	Feb 4, 2020

Abstract

The present disclosure is directed to a method and device for generating anatomical labels for a physiological tree structure. The method may include receiving a 3D model and a 3D skeleton line of the physiological tree structure. The 3D model is restructured based on medical image data of the physiological tree structure acquired by an imaging device. The method further includes selecting at least one level from extracting geometrical features from a pool of selectable levels. The method also includes extracting, by a processor, geometrical features from the 3D model of the physiological tree structure along the 3D skeleton line at the selected at least one level. The method also includes generating, by the processor, anatomical labels for the physiological tree structure using a trained learning network based on the extracted geometrical features.

First claim

What is claimed is:

1. A computer-implemented method for generating anatomical labels for a physiological tree structure, comprising the following steps:
receiving a 3D model and a 3D skeleton line of the physiological tree structure, wherein the 3D model is reconstructed based on medical image data of the physiological tree structure acquired by an imaging device;
selecting at least one level for extracting geometrical features from a pool of selectable levels including a higher level and a lower level, geometrical features in the higher level having a lower granular accuracy than geometrical features in the lower level;
extracting, by a processor, geometrical features from the 3D model of the physiological tree structure along the 3D skeleton line at the selected at least one level; and
generating, by the processor, anatomical labels for the physiological tree structure at a level equal to or higher than the selected level of extracted geometrical features, wherein the anatomical labels are generated using a trained sequence-to-sequence network based on a sequence of the extracted geometrical features.

2. The computer-implemented method of claim 1 , wherein the geometrical features include multiple types of geometrical features at each selected level.

3. The computer-implemented method of claim 1 , wherein the selected at least one level includes a plurality of levels, and wherein extracting geometrical features includes extracting geometrical features at each of the plurality of levels.

4. The computer-implemented method of claim 1 , wherein the selectable levels include levels corresponding to point-wise geometrical features, cell-wise geometrical features, and path-wise geometrical features.

5. The computer-implemented method of claim 1 , wherein the physiological tree structure is a tree structure of any one of blood vessels, tracheal pathway, lymphatic vessels, and nervous tissue.

6. The computer-implemented method of claim 1 , wherein the physiological tree structure is a tree structure of blood vessel, the 3D skeleton line is a centerline.

7. The computer-implemented method of claim 1 , wherein the level for extracting geometrical features is selected based on at least one of the calculating speed, computing resource, and granular accuracy.

8. The computer-implemented method of claim 1 , wherein the trained sequence-to-sequence network is a neural network including a classification layer.

9. The computer-implemented method of claim 8 , wherein the neural network includes at least one of a sequential multi-layer neural network and a tree-structure neural network.

10. The computer-implemented method of claim 9 , wherein generating anatomical labels further includes:
generating a sequence of anatomical labels for the physiological tree structure using a trained sequential LSTM multi-layer neural network or a trained tree-structure LSTM neural network based on the extracted geometrical features.

11. The computer-implemented method of claim 1 , wherein the trained sequence-to-sequence network is trained in an online process or an offline process.

12. The computer-implemented method of claim 1 , further comprising:
mapping the generated anatomical labels to corresponding parts of the physiological tree structure.

13. The computer-implemented method of claim 12 , further comprising:
presenting the generated anatomical labels adjacent to the corresponding parts of the physiological tree structure on a user interface for a user to modify the generated anatomical labels.

14. The computer-implemented method of claim 12 , further comprising:
receiving modified anatomical labels from the user; and
updating the trained sequence-to-sequence network using the extracted geometrical features and the modified anatomical labels as a training dataset.

15. A device for generating anatomical labels for a physiological tree structure, comprising:
an interface configured to receive medical image data of the physiological tree structure acquired by an imaging device; and
a processor configured to:
reconstruct a 3D model and extract a 3D skeleton line of the physiological tree structure based on the received medical image data;
select at least one level for extracting geometrical features from a pool of selectable levels including a higher level and a lower level, geometrical features in the higher level having a lower granular accuracy than geometrical features in the lower level;
extract geometrical features from the 3D model of the physiological tree structure along the 3D skeleton line at the selected at least one level; and
generate anatomical labels for the physiological tree structure at a level equal to or higher than the selected level of the extracted geometrical features, wherein the anatomical labels are generated using a trained sequence-to-sequence network based on a sequence of the extracted geometrical features.

16. The device of claim 15 , wherein the imaging device uses an imaging modality selected from a group of CT, digital subtraction angiography (DSA), MRI, functional MRI, dynamic contrast enhanced MRI, diffusion MRI, spiral CT, cone beam computed tomography (CBCT), positron emission tomography (PET), single-photon emission computed tomography (SPECT), X-ray imaging, optical tomography, fluorescence imaging, ultrasound imaging, radiotherapy portal imaging.

17. A non-transitory computer readable medium having instructions stored thereon, the instructions, when executed by a processor, cause the processor to perform a method for generating anatomical labels for a physiological tree structure, the method comprising:
receiving a 3D model and a 3D skeleton line of the physiological tree structure, wherein the 3D model is restructured based on medical image data of the physiological tree structure acquired by an imaging device;
selecting at least one level for extracting geometrical features from a pool of selectable levels including a higher level and a lower level, geometrical features in the higher level having a lower granular accuracy than geometrical features in the lower level;
extracting geometrical features from the 3D model of the physiological tree structure along the 3D skeleton line at the selected at least one level; and
generating anatomical labels for the physiological tree structure at a level equal to or higher than the selected level of the extracted geometrical features, wherein the anatomical labels are generated using a trained sequence-to-sequence network based on a sequence of the extracted geometrical features.

18. The non-transitory computer readable medium of claim 17 , wherein the selectable levels include levels corresponding to point-wise geometrical features, cell-wise geometrical features, and path-wise geometrical features.

19. The non-transitory computer readable medium of claim 17 , wherein the physiological tree structure is a tree structure of blood vessel, the 3D skeleton line is a centerline.

Assignees

• Beijing Curacloud Tech Co Ltd
• Shenzhen Keya Medical Tech Corporation

Inventors

• Wu Dan
• Wang Xin
• Yin Youbing
• Li Yuwei
• Cao Kunlin
• Song Qi
• Ouyang Bin
• Liang Shuyi

Classifications

• A61B5/489

  Blood vessels · CPC title

• A61B5/743

  Displaying an image simultaneously with additional graphical information, e.g. symbols, charts, function plots · CPC title

• A61B5/055

  involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/30101

  Blood vessel; Artery; Vein; Vascular · CPC title