Automated aorta detection in a CTA volume

Patent metadata

Field	Value
Publication number	US-9691174-B2
Application number	US-201414892086-A
Country	US
Kind code	B2
Filing date	May 26, 2014
Priority date	Jun 5, 2013
Publication date	Jun 27, 2017
Grant date	Jun 27, 2017

Abstract

A method for detecting main body vessels (e.g., the aorta) in a medical volume includes refining the result of a bone removal algorithm.

First claim

The invention claimed is:

1. A method for detecting a main vessel in a volume represented by a digital voxel representation, the method comprising the steps of:
applying a segmentation algorithm to the volume to create a first binary mask with one of a first class value and a second class value assigned to each voxel of the volume;
applying a thresholding operation to the volume to obtain a second binary mask;
subtracting the second binary mask from the first binary mask to generate a third binary mask;
extracting from the third binary mask connected components by propagating labels to all adjacent voxels of the volume;
computing features for each of the connected components; and
preserving the connected components based on results of the step of computing features, and designating a preserved connected component as the vessel.

2. The method according to claim 1 , wherein the step of applying the segmentation algorithm includes the steps of:
subjecting the digital voxel representation to an iterative thresholding operation until a stopping criterion is reached;
finding clusters of the adjacent voxels of the volume by analyzing results of each of the iterative thresholding operations;
building a hierarchical representation of the volume by establishing relations between clusters found in the results of each of the iterative thresholding operations;
assigning a type class to a leaf cluster of the hierarchical representation;
propagating the type class towards a top of the hierarchical representation using propagation rules; and
generating a marking mask to mark locations of voxels of a specific class by merging the locations of voxels contained in clusters that received that type class through the step of propagating.

3. The method according to claim 1 , wherein the first class values are assigned to voxels of osseous tissue and the second class values are assigned to voxels of vascular tissue.

4. The method according to claim 1 , wherein the volume is obtained by a computed tomography angiography procedure and the vessel is the aorta.

5. The method according to claim 1 , wherein the features are at least one of a number of voxels in the connected component and a shape of the connected component.

6. The method according to claim 2 , wherein the marking mask marks the locations of voxels that were contained in top ancestral clusters.

7. The method according to claim 2 , wherein the stopping criterion is met when no clusters are generated that fulfill a minimum size requirement.

8. The method according to claim 2 , wherein the stopping criterion is reached when a threshold is below a given limit value.

9. The method according to claim 2 , wherein a class to be assigned to one of the clusters of the adjacent voxels of the volume is determined based on results of an analysis of values of qualitative and/or quantitative features determined for the one of the clusters.

10. The method according to claim 2 , wherein a class to be assigned to a leaf is decided upon by a trained classifier.

11. The method according to claim 2 , further comprising performing a first type of post processing including adding voxels to the marking mask to restore voxels lost during the iterative thresholding operation.

12. The method according to claim 11 , wherein the first type of post processing includes a distance transform-based assignment process.

13. A non-transitory computer readable medium comprising a computer executable program code adapted to carry out the steps of claim 1 when the computer executable program code is executed on a computer.

Assignees

• Agfa Healthcare
• Agfa Healthcare Nv

Inventors

• Ouji Asma
• Vergote Koen

Classifications

• G06V40/14

  Vascular patterns · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06T7/11 Primary

  Region-based segmentation · CPC title

• G06T15/08 Primary

  Volume rendering · CPC title

• G06F18/23

  Clustering techniques · CPC title