Three-dimensional medical image multi-organ reconstruction method based on context learning
By adopting a three-dimensional medical imaging multi-organ reconstruction method based on context learning in medical imaging segmentation, the global context information and feature multi-scale cross-fusion module of the support set are used to solve the problem of accuracy reduction of existing methods under new tasks and new feature distributions, and efficient multi-organ reconstruction and context reasoning are achieved.
Patent Information
- Application Number
- CN202411727364.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-28
- Publication Date
- 2025-03-25
- Estimated Expiration
- 2044-11-28
AI Technical Summary
Existing medical imaging segmentation methods work with new anatomical structures, image modalities, or tasks with unknown labels, the model cannot effectively adapt to the new feature distribution, resulting in reduced accuracy and requires a lot of computing resources and expertise.
A three-dimensional medical imaging multi-organ reconstruction method based on context learning is adopted, and the global context information provided by the support set is used to integrate information across space through feature multi-scale cross-fusion module and encoder-decoder structure to improve the generalization ability and prediction accuracy of the model.
Multi-organ reconstruction of downstream domain three-dimensional medical images can be achieved without training, which improves the learning efficiency and contextual reasoning capabilities of the network, and reduces the dependence on computing resources and professional knowledge.