Retriever model training method, birt-hogg-dube syndrome identification method and system based on retrieval enhancement generation

By constructing positive and negative sample pairs and using an adaptive metric variance mechanism to adjust the angle margin, the retrieval model's category decision boundary in cosine space is enhanced, solving the diagnostic problem caused by the similarity of BHD and DCLD image features and achieving more accurate image recognition.

CN121366329BActive Publication Date: 2026-05-05UNIV OF SCI & TECH OF CHINA
2 Cites 0 Cited by

Patent Information

Application Number
CN202511535073.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-10-24
Publication Date
2026-05-05
Estimated Expiration
2045-10-24

AI Technical Summary

Technical Problem

Because Bert-Hogg-Dubb syndrome (BHD) has similar features to other diffuse cystic lung lesions (DCLDs) on CT images, diagnosis is difficult. Existing deep learning models are prone to hallucinations when identifying these lesions, making accurate differentiation difficult.

Method used

By constructing positive and negative sample pairs and using an adaptive metric variance mechanism to adjust the angle margin, the retrieval model's category decision boundary in cosine space is enhanced. Combined with multi-stage downsampling and attention modules, the recognition process of large models is optimized.

Benefits of technology

This improved the accuracy of large-scale model recognition of BHD images, reduced the false diagnosis rate, and ensured the accuracy and rationality of the output results.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

This application discloses a retrieval model training method, a retrieval-enhanced generation-based Bert-Hogg-Dubbs syndrome (BHD) identification method and system, relating to the field of rare disease identification. The training method includes acquiring positive and negative sample pairs; inputting the positive and negative sample pairs into an initial retrieval model; calculating the loss function of the initial retrieval model; and dynamically adjusting the angle margin of the loss function in real time according to a metric variance adaptive mechanism. The initial retrieval model is then optimized based on the loss function calculation results. This application forcibly expands the angle interval between BHD and non-BHD by using the angle margin of the loss function, and uses a metric variance adaptive mechanism to dynamically adjust the angle margin based on the statistical variance of the cosine similarity among all positive sample pairs in the current training batch. This solves the problem of weak image differences and blurred category decision boundaries in DCLDs caused by the highly similar imaging features of various rare diseases, thus improving the recognition accuracy of large models for query information.
Need to check novelty before this filing date? Find Prior Art