Pathological image classification system and method based on Mama and hybrid expert model
By combining Mamba and hybrid expert models, a new pathological image classification system was designed, which solved the problem of high computational cost of Transformer when dealing with high-resolution WSI, achieved efficient feature representation and package label prediction, and improved the classification performance of WSI tasks.
CN119851017AActive Publication Date: 2025-04-18HARBIN INST OF TECH
6 Cites 3 Cited by
Patent Information
- Application Number
- CN202411913875.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-24
- Publication Date
- 2025-04-18
- Estimated Expiration
- 2044-12-24
AI Technical Summary
Technical Problem
When handling high-resolution WSI, the secondary complexity problem of Transformer leads to high computational costs and it is difficult to efficiently process large-scale WSI data.
Method used
A pathological image classification system based on Mamba and hybrid expert model is proposed. By introducing two-way Mamba blocks, dynamic memory banks and cross-pseudo-packet fusion modules, combined with expert hybrid modules, we can realize efficient feature representation and package label prediction.
Benefits of technology
It effectively reduces computing costs, improves the classification performance of WSI tasks, and can better process high-resolution WSI data, and achieves finer-grained packet representation and more robust packet label prediction.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a pathological image classification system and method based on Mama and a hybrid expert model, the pathological image classification system proposes a challenging task for WSI classification by combining the advantages of the Mama and the hybrid expert model, and develops a novel Mama-based hierarchical network called Mama-WSI. In order to promote feature representation while ensuring efficiency, the Mama-WSI introduces bidirectional Mama blocks in a pseudo packet and across the pseudo packet to explore correlation between instances in the pseudo packet and global pseudo packet representation. Meanwhile, a dynamic memory library is provided to store historical information related to label prediction, and a cross-pseudo packet fusion module is designed to fuse effective information from a dynamically updated dictionary module so as to enhance current pseudo packet representation and global modeling. Furthermore, the expert hybrid module is capable of adaptively coordinating global and local packet representations to achieve robust packet label prediction.
Need to check novelty before this filing date? Find Prior Art