Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Multi-Channel Network Predictive Optimization Method for Coronary Segmentation

A technology of network prediction and optimization methods, which is applied in image analysis, image enhancement, instruments, etc., can solve problems such as not being robust, and achieve the effect of increasing learning ability

Active Publication Date: 2021-07-13
数坤(上海)医疗科技有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Fractured segmentation results are not robust to subsequent image processing

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0018] This embodiment discloses a multi-channel network prediction optimization method for coronary artery segmentation, which includes the following steps:

[0019] S1. Using the CT image + ignoring the edge features in the Z direction as a sample, train the model Z;

[0020] S2. Input the CT image to be segmented and its edge features ignoring the Z direction into the model Z to obtain a first segmentation result.

Embodiment 2

[0022] S1. Using CT images and edge features in X, Y, and Z directions as samples, train model A.

[0023] S2. Input the CT image to be segmented and its edge features in the X, Y, and Z directions into the model A to obtain a first segmentation result.

[0024] Comparing the segmentation result of this embodiment with the segmentation result of Example 1, both models use CELoss as the loss function, and the final loss result: Example 1 is 0.018, and Example 2 is 0.03. It can be seen that the implementation The result precision obtained by the method adopted in Example 1 is obviously better than that in Example 2.

[0025] At the same time, the centerlines were extracted from the segmentation results of Example 1 and the segmentation results of Example 2, respectively, and the breakpoint detection was performed on the centerlines of the two respectively. In the final result, the number of breakpoints in Example 2 was significantly more than that in Example 1. of breaking poin...

Embodiment 3

[0027] Although the method of Embodiment 1 is used for segmentation, the accuracy is greatly improved, but there are still breakpoints unavoidably. Therefore, the present invention also repairs subsequent breakpoints.

[0028] S3. Using CT images + ignoring the edge features in the X direction and CT images + ignoring the edge features in the Y direction as samples, train the X model and the Y model respectively

[0029] S4. Perform centerline extraction on the first segmentation result;

[0030] S5. Performing break point detection on the extracted center line;

[0031] S6. Give an estimated direction to the breaking point according to the center point of the preceding sequence, and calculate the normal direction of the estimated direction;

[0032] S7. Calculate the components of the X direction and the Y direction in the normal direction of the estimated direction, take the direction with a large component as the main direction, and use the main direction as the neglected ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a multi-channel network prediction optimization method for coronary artery segmentation, comprising: S1, using CT images + ignoring the edge features in the Z direction as samples, training model Z; S2, taking the CT image to be segmented and its ignoring The edge features in the Z direction are input to the model Z to obtain the first segmentation result; the CT image+ignoring the edge features in the X direction and the CT image+ignoring the edge features in the Y direction are used as samples to train the X model and the Y model; S4. Extract the center line from the segmentation result; S5, detect the breakpoint of the extracted center line; S6, give the estimated direction of the break point, and calculate the normal direction of the estimated direction; S7, calculate the distance between the X direction and the Y direction For the component in the normal direction, take the direction with the larger component as the ignored direction, extract the CT image to be segmented and the edge corresponding to the direction and input the corresponding model to obtain the second segmentation result; S8, extract the segmentation at the breaking point from the second segmentation result, replace The corresponding position in the first segmentation result.

Description

technical field [0001] The invention relates to the technical field of coronary artery image post-processing, in particular to a multi-channel network prediction and optimization method for coronary artery segmentation. Background technique [0002] Automated coronary reconstruction has important clinical value and practical significance for doctors. Since the trend of the coronary artery is mostly along the Z direction (scanning direction), the coronary artery often breaks during automatic segmentation. Fractured segmentation results are not robust to subsequent image processing steps. Contents of the invention [0003] The purpose of the present invention is to provide a multi-channel network prediction and optimization method for coronary artery segmentation. [0004] To achieve the above object, the present invention adopts the following technical solutions: [0005] A multi-channel network predictive optimization method for coronary artery segmentation, including: ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06T7/12G06T7/181
CPCG06T2207/10081G06T2207/20081G06T2207/20084G06T2207/30004G06T7/12G06T7/181
Inventor 肖月庭阳光郑超
Owner 数坤(上海)医疗科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com