Model-based image segmentation

Patent metadata

Field	Value
Publication number	US-12062185-B2
Application number	US-202117797786-A
Country	US
Kind code	B2
Filing date	Feb 5, 2021
Priority date	Feb 14, 2020
Publication date	Aug 13, 2024
Grant date	Aug 13, 2024

Abstract

A method and system for mapping boundary detecting features of at least one source triangulated mesh of known topology to a target triangulated mesh of arbitrary topology. A region of interest in a volumetric image associated with each triangle of the target triangulated mesh is provided to a feature mapping network. The feature mapping network assigns a feature selection vector to each triangle of the target triangulated mesh. The associated region of interest and assigned feature selection vector for each triangle of the target triangulated mesh are provided to a boundary detection network. A predicted boundary based on features of the associated region of interest selected by the assigned feature selection vector is obtained from the boundary detection network.

First claim

The invention claimed is:

1. A computer-implemented method of predicting a boundary of an object in a region of interest, the method being suitable for mapping boundary detecting features of at least one source triangulated mesh of known topology to a target triangulated mesh of arbitrary topology, the computer-implemented method comprising:
providing, to a feature mapping network, the region of interest in a volumetric image associated with each triangle of the target triangulated mesh, wherein the region of interest comprises each triangle and part of a boundary of the object that the target triangulated mesh is to delineate and a part of the object's surroundings, the feature mapping network is configured to assign a feature selection vector to each triangle of the target triangulated mesh based on the region of interest for selecting thereby mapping the boundary detecting features of a triangle of the at least one source triangulated mesh to the triangle of the target triangulated mesh;
assigning the feature selection vector to each triangle of the target triangulated mesh using the feature mapping network;
for each triangle of the target triangulated mesh, providing the associated region of interest and assigned feature selection vector to a boundary detection network, wherein the boundary detection network is configured to detect a predicted boundary for each triangle of the target triangulated mesh based on features of the region of interest selected by the assigned feature selection vector; and
for each triangle of the target triangulated mesh, obtaining a predicted boundary from the boundary detection network based on features of the associated region of interest selected by the assigned feature selection vector.

2. The computer-implemented method of claim 1 , further comprising obtaining the target triangulated mesh of arbitrary topology by adding triangles to or removing triangles from a triangulated mesh of known topology.

3. The computer-implemented method of claim 1 , further comprising obtaining the target triangulated mesh of arbitrary topology by:
obtaining a segmentation of the volumetric image using a voxel-wise segmentation technique; and
generating the target triangulated mesh based on the obtained segmentation.

4. The computer-implemented method of claim 1 , wherein the region of interest associated with each triangle is oriented according to the normal of the triangle.

5. The computer-implemented method of claim 1 , wherein the feature mapping network is trained using a first training algorithm configured to receive an array of training inputs and known outputs, wherein the training inputs comprise regions of interest associated with triangles of meshes of arbitrary topology and the known outputs comprise known boundaries for the regions of interest.

6. The computer-implemented method of claim 5 , wherein the first training algorithm is further configured to:
assign a feature selection vector for each region of interest associated with a triangle of a mesh of arbitrary topology;
for each triangle of a mesh of arbitrary topology, provide the associated region of interest and the assigned feature selection vector to the boundary detection network;
for each triangle of a mesh of arbitrary topology, obtain a predicted boundary from the boundary detection network; and
train weights of the feature mapping network based on the predicted boundaries from the boundary detection network and the known boundaries.

7. The computer-implemented method of claim 1 , further comprising providing a set of standardized coordinates for each region of interest to the feature mapping network.

8. The computer-implemented method of claim 1 , further comprising providing a relative position of each triangle of the target triangulated mesh to the feature mapping network.

9. The computer-implemented method of claim 1 , wherein the boundary detection network is trained using a second training algorithm configured to receive an array of training inputs and known outputs, wherein the training inputs comprise training images with meshes of known topology and the known outputs comprise known boundaries.

10. A computer-implemented method of model-based image segmentation, comprising:
mapping features of at least one source triangulated mesh of known topology to a target triangulated mesh of arbitrary topology, according to the method of claim 1 ; and
segmenting an object from the volumetric image using the target triangulated mesh.

11. A system for predicting a boundary of an object in a region of interest of an object, comprising:
a processor; and
a memory storing a computer program, wherein the processor when executing the computer program the processor is caused to:
provide, to a feature mapping network, the region of interest in a volumetric image associated with each triangle of a target triangulated mesh wherein the region of interest comprises each triangle and part of a boundary of the object that the target triangulated mesh is to delineate and a part of the object's surroundings, the feature mapping network is configured to assign a feature selection vector to each triangle of the target triangulated mesh based on the region of interest for selecting thereby mapping the boundary detecting features of a triangle of the at least one source triangulated mesh to the triangle of the target triangulated mesh;
assign the feature selection vector to each triangle of the target triangulated mesh using the feature mapping network;
for each triangle of the target triangulated mesh, provide the associated region of interest and assigned feature selection vector to a boundary detection network, wherein the boundary detection network is configured to detect a predicted boundary for each triangle of the target triangulated mesh based on features of the region of interest selected by the assigned feature selection vector; and
for each triangle of the target triangulated mesh, obtain a predicted boundary from the boundary detection network based on features of the associated region of interest selected by the assigned feature selection vector.

12. The system of claim 11 , further adapted to orient the region of interest associated with each triangle according to the normal of the triangle.

13. The system of claim 11 , further adapted to provide, for each triangle of the target triangulated mesh, at least one of a set of standardized coordinates of the associated region of interest and a relative position of the triangle to the feature mapping network.

14. A model-based image segmentation system, comprising:
the system of claim 11 , further adapted to segment an object from the volumetric image using the target triangulated mesh; and
a user interface configured to receive, from the system, and display the segmented image of the object.

15. A computer-implemented method of predicting a boundary of an object in a region of interest, the method being suitable for mapping boundary detecting features of at least one source triangulated mesh of known topology to a target triangulated mesh of arbitrary topology, the computer-implemented method comprising:
providing, to a feature mapping network, the regions of interest in a volumetric image respectively associated with triangles of the target triangulated mesh, wherein the region of interest comprises at least one of the triangles and part of a boundary of the object that the target triangulated mesh is to delineate and a part of the object's surroundings, the feature mapping network is configured to respectively assign a feature selection vector to the triangles of the target

Assignees

• Koninklijke Philips Nv

Inventors

• Buerger Christian
• Klinder Tobias
• Schmidt-Richberg Alexander

Classifications

• G06T2207/30004

  Biomedical image processing · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2200/24

  involving graphical user interfaces [GUIs] · CPC title

• G06T17/205

  Re-meshing · CPC title