Multiple Instance Learning Framework with Masked Hard Instance Mining for Gigapixel Histopathology Image Analysis

2025-09-14

AI Summary

To address the challenge of sparse positive instances in gigapixel histopathological images—which causes conventional multiple instance learning (MIL) models to overlook hard examples and yield inaccurate boundary modeling—this paper proposes Mask-based Hard Instance Mining (MHIM). MHIM employs a momentum teacher–student framework that dynamically masks salient instances via class-aware instance probabilities and large-scale random masking, implicitly steering attention toward hard-to-classify regions. It further incorporates a global recovery network and consistency regularization to ensure stable and diverse hard instance discovery. Technically, MHIM integrates a Siamese architecture, exponential moving average parameter updates, and end-to-end optimization. Evaluated across 12 benchmark tasks—including cancer diagnosis, subtyping, and survival analysis—MHIM consistently outperforms state-of-the-art methods, delivering significant improvements in discriminative performance and inference efficiency. The source code is publicly available.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Segmentation
• Search and Optimization: Mixed Discrete/Continuous Search

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Digitizing pathological images into gigapixel Whole Slide Images (WSIs) has opened new avenues for Computational Pathology (CPath). As positive tissue comprises only a small fraction of gigapixel WSIs, existing Multiple Instance Learning (MIL) methods typically focus on identifying salient instances via attention mechanisms. However, this leads to a bias towards easy-to-classify instances while neglecting challenging ones. Recent studies have shown that hard examples are crucial for accurately modeling discriminative boundaries. Applying such an idea at the instance level, we elaborate a novel MIL framework with masked hard instance mining (MHIM-MIL), which utilizes a Siamese structure with a consistency constraint to explore the hard instances. Using a class-aware instance probability, MHIM-MIL employs a momentum teacher to mask salient instances and implicitly mine hard instances for training the student model. To obtain diverse, non-redundant hard instances, we adopt large-scale random masking while utilizing a global recycle network to mitigate the risk of losing key features. Furthermore, the student updates the teacher using an exponential moving average, which identifies new hard instances for subsequent training iterations and stabilizes optimization. Experimental results on cancer diagnosis, subtyping, survival analysis tasks, and 12 benchmarks demonstrate that MHIM-MIL outperforms the latest methods in both performance and efficiency. The code is available at: https://github.com/DearCaat/MHIM-MIL.

Problem

• Mining hard instances in gigapixel pathology images
• Overcoming bias toward easy-to-classify tissue instances
• Improving multiple instance learning for cancer analysis

Innovation

• Masked hard instance mining for histopathology
• Siamese structure with consistency constraint
• Momentum teacher masks salient instances