Fast and Accurate Gigapixel Pathological Image Classification with Hierarchical Distillation Multi-Instance Learning

2025-02-28

AI Summary

To address the high inference cost of multi-instance learning (MIL) in gigapixel whole-slide image (WSI) classification, this paper proposes Hierarchical Distillation MIL (HDMIL). HDMIL introduces a novel collaborative architecture comprising a Dynamic Multi-Instance Network (DMIN) and a lightweight Instance Pre-screening Network (LIPN), enabling low-resolution coarse screening to guide high-resolution fine-grained inference. It further incorporates attention mask distillation and a Kolmogorov–Arnold network classifier parameterized by learnable Chebyshev polynomials. On Camelyon16, HDMIL achieves a 3.13% AUC improvement and 28.6% faster inference; it consistently outperforms state-of-the-art methods across three public benchmarks. Key contributions include: (1) the first hierarchical distillation MIL paradigm for WSIs; (2) the first Chebyshev-driven, learnable activation-based classifier; and (3) an efficient block-level dynamic instance selection mechanism.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Multiagent Systems: Multiagent Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Although multi-instance learning (MIL) has succeeded in pathological image classification, it faces the challenge of high inference costs due to processing numerous patches from gigapixel whole slide images (WSIs). To address this, we propose HDMIL, a hierarchical distillation multi-instance learning framework that achieves fast and accurate classification by eliminating irrelevant patches. HDMIL consists of two key components: the dynamic multi-instance network (DMIN) and the lightweight instance pre-screening network (LIPN). DMIN operates on high-resolution WSIs, while LIPN operates on the corresponding low-resolution counterparts. During training, DMIN are trained for WSI classification while generating attention-score-based masks that indicate irrelevant patches. These masks then guide the training of LIPN to predict the relevance of each low-resolution patch. During testing, LIPN first determines the useful regions within low-resolution WSIs, which indirectly enables us to eliminate irrelevant regions in high-resolution WSIs, thereby reducing inference time without causing performance degradation. In addition, we further design the first Chebyshev-polynomials-based Kolmogorov-Arnold classifier in computational pathology, which enhances the performance of HDMIL through learnable activation layers. Extensive experiments on three public datasets demonstrate that HDMIL outperforms previous state-of-the-art methods, e.g., achieving improvements of 3.13% in AUC while reducing inference time by 28.6% on the Camelyon16 dataset.

Problem

• Reduces high inference costs in gigapixel pathological image classification.
• Eliminates irrelevant patches to improve classification speed and accuracy.
• Introduces a novel classifier enhancing performance in computational pathology.

Innovation

• Hierarchical distillation for efficient patch elimination
• Dynamic and lightweight networks for multi-resolution analysis
• Chebyshev-polynomials-based classifier enhancing performance