AINet: Anchor Instances Learning for Regional Heterogeneity in Whole Slide Image

2026-02-21

AI Summary

This work addresses the challenge of regional heterogeneity in whole-slide images caused by the sparsity and morphological diversity of tumors. To this end, we propose an Anchor Instance (AI) mechanism and develop an efficient multiple instance learning framework, termed AINet. Our approach employs a Dual-level Anchor Mining (DAM) module to identify key instances that are both locally representative and globally discriminative, and introduces an Anchor-guided Region Correction (ARC) module to integrate complementary information across regions, thereby refining regional representations. Both DAM and ARC are designed as modular, plug-and-play components that preserve regional diversity while suppressing non-discriminative patterns, significantly reducing computational overhead. Experiments demonstrate that AINet achieves superior performance over state-of-the-art methods with fewer parameters and FLOPs, and that DAM and ARC can be generically integrated to enhance mainstream MIL frameworks.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Recent advances in multi-instance learning (MIL) have witnessed impressive performance in whole slide image (WSI) analysis. However, the inherent sparsity of tumors and their morphological diversity lead to obvious heterogeneity across regions, posing significant challenges in aggregating high-quality and discriminative representations. To address this, we introduce a novel concept of anchor instance (AI), a compact subset of instances that are representative within their regions (local) and discriminative at the bag (global) level. These AIs act as semantic references to guide interactions across regions, correcting non-discriminative patterns while preserving regional diversity. Specifically, we propose a dual-level anchor mining (DAM) module to \textbf{select} AIs from massive instances, where the most informative AI in each region is extracted by assessing its similarity to both local and global embeddings. Furthermore, to ensure completeness and diversity, we devise an anchor-guided region correction (ARC) module that explores the complementary information from all regions to \textbf{correct} each regional representation. Building upon DAM and ARC, we develop a concise yet effective framework, AINet, which employs a simple predictor and surpasses state-of-the-art methods with substantially fewer FLOPs and parameters. Moreover, both DAM and ARC are modular and can be seamlessly integrated into existing MIL frameworks, consistently improving their performance.

Problem

• whole slide image
• regional heterogeneity
• multi-instance learning
• tumor sparsity
• morphological diversity

Innovation

• anchor instance
• regional heterogeneity
• multi-instance learning
• whole slide image
• modular correction