Weakly supervised pathological image tissue segmentation method based on text prompt learning

By employing text-based learning and multimodal interaction optimization, the problems of tissue co-occurrence and blurred boundaries in pathological images were solved, achieving efficient and accurate tissue segmentation in pathological images, reducing annotation costs, and improving the applicability of the model.

CN121458971APending Publication Date: 2026-02-03GUILIN UNIV OF ELECTRONIC TECH
0 Cites 2 Cited by

Patent Information

Application Number
CN202511528677.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-10-24
Publication Date
2026-02-03

AI Technical Summary

Technical Problem

Existing supervised learning-based pathological image segmentation methods rely on a large number of pixel-level annotations, resulting in high labor costs and difficulty in application in actual clinical diagnosis. Furthermore, weakly supervised methods based on CAM are inaccurate in segmentation when dealing with tissue co-occurrence and blurred boundaries.

Method used

A text-based learning approach is adopted, which uses pre-trained text encoders and image encoders to extract features, generates an initial class activation map through a cross-attention mechanism, optimizes the boundary using a multimodal context optimization module (MCRM), and trains a segmentation model using a hybrid loss function to generate a refined pseudomask for segmentation.

Benefits of technology

It reduces the reliance on pixel-level annotation, improves the accuracy of tissue segmentation in pathological images and the generalization ability of the model, reduces annotation costs, and enhances the segmentation performance of the model in the absence of pixel-level supervision.

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

Abstract

The invention discloses a weak supervision pathological image tissue segmentation method based on text prompt learning. The method comprises the steps of feature extraction and initial class activation graph generation; using an MCRM module to optimize the initial class activation graph to obtain a refined class activation graph; and aggregating the plurality of refined class activation graphs to form a fused pseudo mask, taking the fused pseudo mask as a supervision signal, training a segmentation model, and after the training is completed, segmenting the new pathological image tissue by using the segmentation model. According to the method, a text prompt learning mechanism is utilized to focus the model on learning high-discrimination features, so that the influence of tissue co-occurrence is reduced. An initial class activation graph is optimized through a multi-mode class activation graph refining module, and the integrity of boundary segmentation is enhanced. Meanwhile, pseudo masks from different network layers are fused to train a segmentation model, and semantic segmentation of the pathological image is realized. According to the method, high-annotation data dependence is effectively relieved, and the generalization ability of the model is improved, so that application in the field of artificial intelligence-assisted medical treatment is promoted.
Need to check novelty before this filing date? Find Prior Art