一种磁共振引导的目标区域边界提取系统及方法
By generating coarse boundaries through compression and mapping relationships, and combining global information integration and local feature alignment, the problem of balancing speed and accuracy in target boundary extraction in existing technologies is solved, achieving efficient and accurate target boundary extraction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SICHUAN CANCER HOSPITAL
- Filing Date
- 2026-04-28
- Publication Date
- 2026-07-17
AI Technical Summary
Existing target boundary extraction systems in MRI-guided radiotherapy struggle to balance the processing speed of high-resolution dynamic images with the accuracy of target delineation. Directly reducing the resolution of the input image leads to the loss of key anatomical details in the transition zones of small blood vessel edges and organ boundaries. When performing fine segmentation only on local high-resolution areas, the lack of global spatial context guidance makes it difficult to accurately determine the relative positional relationship between the target area and surrounding tissues.
The original high-resolution 3D image is compressed into low-resolution compressed video information by the compressed video information acquisition module. The mapping relationship is obtained by the inverse feature extraction module. The coarse boundary is generated by the first boundary extraction model. The supplementary information module extracts local high-resolution regions. The second boundary extraction model performs global information integration and feature alignment to achieve accurate extraction of the target area boundary.
It significantly improves the speed of target boundary extraction while preserving key anatomical details, ensuring accurate determination of the relative positional relationship between the target area and surrounding tissues, and achieving the optimal balance between target boundary extraction speed and accuracy.
Smart Images
Figure CN122134751B_ABST