# US 2021/0279906 A1 · Method, device and system for generating a centerline for an object in an image

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2021/0279906 A1
- Document ID: US20210279906A1
- Status: Published application
- Kind: A1
- Application number: 17/330,557
- Filed: 2021-05-26
- Published: 2021-09-09
- Earliest claimed: 2018-08-23
- Assignees: Shenzhen Keya Medical Technology Corporation
- Applicants: Shenzhen Keya Medical Technology Corporation
- Inventors: Junjie Bai; Zhihui Guo; Youbing Yin; Xin Wang; Yi Lu; Kunlin Cao; Qi Song; Xiaoyang Xu; Bin Ouyang
- Web page: https://patentgrep.com/patent/US20210279906A1

## CPC

- G06T 7/11: Region-based segmentation
- G06T 7/68: of symmetry
- G06N 3/045: Combinations of networks
- G06N 3/0455: Auto-encoder networks; Encoder-decoder networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/08: Learning methods
- G06N 3/09: Supervised learning
- G06N 20/20: Ensemble learning
- G06T 2207/10072: Tomographic images
- G06T 2207/10116: X-ray image
- G06T 2207/20041: Distance transform
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30101: Blood vessel; Artery; Vein; Vascular
- G06T 2207/30172: Centreline of tubular or elongated structure
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

Systems and methods for generating a centerline for an object in an image are provided. An exemplary method includes receiving an image containing the object. The method also includes detecting at least one bifurcation of the object using a trained bifurcation learning network based on the image. The method further includes extracting the centerline of the object based on a constraint condition that the centerline passes through the detected bifurcation.

## Claims

### Claim 1

A method for generating a centerline of an object, comprising:
receiving an image containing the object, wherein the image is acquired by an imaging device;
detecting, by a processor, at least one bifurcation of the object using a trained bifurcation learning network based on the image; and
extracting, by the processor, the centerline of the object based on a constraint condition that the centerline passes through the detected bifurcation.

### Claim 2 (depends on 1)

The method of claim 1, wherein the method further comprises generating, by the processor, a distance cost image using a trained distance cost learning network based on the image, wherein the extracting is performed further based on the distance cost image.

### Claim 3 (depends on 2)

The method of claim 2, wherein an intensity of each pixel of the distance cost image is indicative of a distance of the pixel from the centerline of the object.

### Claim 4 (depends on 1)

The method of claim 1, further comprising detecting, by the processor, end points of the object using a trained end point learning network based on the image, wherein the extracting is performed further based on the end points.

### Claim 5 (depends on 4)

The method of claim 4, wherein the detecting the end points of the object further comprises:
predicting, by the processor, an end point map using the end point learning network based on the image, wherein an intensity of a pixel of the end point map is indicative of whether the pixel corresponds to a starting end point or a terminating end point, or the pixel does not correspond to an ending point.

### Claim 6 (depends on 4)

The method of claim 4, wherein the extracting the centerline of the object comprises:
generating a path connecting a pair of end points with a minimal distance cost, wherein the path passes through the detected bifurcation.

### Claim 7 (depends on 1)

The method of claim 1, wherein the bifurcation learning network is a convolutional network.

### Claim 8 (depends on 1)

The method of claim 1, wherein the extracted centerline of the object comprises a single pixel-wide line.

### Claim 9 (depends on 1)

The method of claim 1, wherein the bifurcation learning network comprises an encoder configured to extract features, and the encoder further comprises an attention unit applying larger weights on features extracted at a location corresponding to the object compared to weights on features extracted at another location not corresponding to the object.

### Claim 10 (depends on 1)

The method of claim 1, wherein the bifurcation learning network comprises an encoder configured to extract features and a decoder configured to detect the bifurcations of the object based on the extracted features.

### Claim 11 (depends on 1)

The method of claim 1, wherein the object comprises a vessel.

### Claim 12

A system for generating a centerline of an object, comprising:
an interface configured to receive an image containing the object, wherein the image is acquired by an imaging device; and
a processor configured to:
detect at least one bifurcation of the object using a trained bifurcation learning network based on the image; and
extract the centerline of the object based on a constraint condition that the centerline passes through the detected bifurcation.

### Claim 13 (depends on 12)

The system of claim 12, wherein the processor is further configured to generate a distance cost image using a trained distance cost learning network based on the image, wherein the extracting is performed further based on the distance cost image.

### Claim 14 (depends on 12)

The system of claim 12, wherein the processor is further configured to detect end points of the object using a trained end point learning network based on the image, wherein the extracting is performed further based on the end points.

### Claim 15 (depends on 14)

The system of claim 14, wherein the processor is further configured to:
generate a path connecting a pair of a starting end point and a corresponding terminating end point with a minimal distance cost, wherein the path passes through the detected bifurcation.

### Claim 16 (depends on 12)

The system of claim 12, wherein the bifurcation learning network is a convolutional network.

### Claim 17 (depends on 12)

The system of claim 12, wherein the extracted centerline of the object comprises a single pixel-wide line

### Claim 18 (depends on 12)

The system of claim 12, wherein the bifurcation learning network comprises an encoder configured to extract features, and the encoder further comprises an attention unit applying larger weights on features extracted at a location corresponding to the object compared to weights on features extracted at another location not corresponding to the object.

### Claim 19 (depends on 12)

The system of claim 12, wherein the object comprises a vessel.

### Claim 20

A non-transitory computer readable medium storing instructions that, when executed by a processor, perform a method for generating a centerline of an object, the method comprising:
receiving an image containing the object, wherein the image is acquired by an imaging device;
detecting at least one bifurcation of the object using a trained bifurcation learning network based on the image; and
extracting the centerline of the object based on a constraint condition that the centerline passes through the detected bifurcation.
