一种基于双向跨模态交互的放射学报告生成方法
By employing a bidirectional cross-modal interactive radiology report generation method, which utilizes image encoding and knowledge aggregation modules to achieve bidirectional interaction between image features and knowledge features, the quality of generated radiology reports is improved, and the semantic consistency and sentence similarity of the reports are enhanced.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES)
- Filing Date
- 2026-05-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing radiology report generation methods lack bidirectional cross-modal interaction between image features and knowledge features, resulting in significant differences between the generated radiology reports and actual clinical radiology reports.
A radiology report generation method based on bidirectional cross-modal interaction is adopted. The radiology report generation network model is used for image encoding, knowledge aggregation and bidirectional cross-attention fusion to realize bidirectional cross-modal interaction and collaborative optimization between image features and knowledge features.
The generated radiology report is closer to the sample radiology report text in terms of sentence structure, semantic coherence and consistency, and has a higher sentence similarity.
Smart Images
Figure CN122245596B_ABST