A blood vessel segmentation method and device
By constructing a vascular distribution map and fusion of multi-scale features across networks, the problem of omission or fracture of vascular segmentation results in the prior art is solved, and higher vascular segmentation accuracy and ability to adapt to complex scenarios are achieved.
CN113554665BActive Publication Date: 2025-03-28HANGZHOU SHENRUI BOLIAN TECH CO LTD +1
2 Cites 0 Cited by
Patent Information
- Application Number
- CN202110768617.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-07-07
- Publication Date
- 2025-03-28
- Estimated Expiration
- 2041-07-07
AI Technical Summary
Technical Problem
The existing vascular segmentation methods cannot effectively deal with artifacts and noise caused by external factors, resulting in missing or breaking of the segmentation results.
Method used
The vascular pre-segmentation method based on multi-scale feature extraction is adopted, and global feature modeling is carried out by constructing a vascular distribution map, combined with multi-scale feature fusion across networks, to achieve the prediction of vascular regions.
Benefits of technology
It improves the accuracy of vascular segmentation, can better adapt to vascular structures in complex scenarios, and avoids the model predicting segmentation areas that do not conform to vascular structures.
✦ Generated by Eureka AI based on patent content.
Abstract
The present invention provides a blood vessel segmentation method and device. The method comprises: inputting a CTA image into a first convolutional network to perform blood vessel pre-segmentation based on multi-scale feature extraction; constructing a blood vessel distribution map; inputting the CTA image into a second convolutional network, and multiplying the output of the second convolutional network with the input CTA image and inputting it into a third convolutional network; the third convolutional network and the graph convolutional network perform feature interaction based on bidirectional mapping, perform global feature modeling based on the blood vessel distribution map to capture multi-scale local and global features, and achieve blood vessel region prediction; and predict blood vessel segmentation results by fusing multi-scale features. The present invention performs blood vessel segmentation based on cross-network multi-scale feature fusion and local feature and global feature fusion. Compared with the prior art that performs blood vessel segmentation based on local features, the present invention can effectively avoid the model predicting segmented areas that do not conform to the blood vessel structure, thereby improving the accuracy of blood vessel segmentation.
Need to check novelty before this filing date? Find Prior Art