Anatomy-Structured Hierarchical MIL for Weakly-Supervised Thoracic Disease Detection in Chest X-rays

2026-09-27

AI Summary

This study addresses the localization challenges in weakly supervised chest X-ray disease detection caused by subtle lesions and anatomical overlap. We propose the ASH-MIL framework, which operates without bounding box annotations. By injecting soft spatial priors through parallel anatomy-structured branches, our method enables dynamic instance discovery. Furthermore, it integrates hierarchical multiple instance learning, cross-attention mechanisms, and anatomy-aware spatial biases to generate evidence maps and achieve precise lesion localization. Extensive experiments on the CXR8 and MIMIC-CXR datasets demonstrate that the proposed framework significantly outperforms existing methods under rigorous localization metrics.

Abstract

Weakly-supervised thoracic disease detection in chest X-rays (CXR) is challenging due to subtle appearances and complex anatomical overlap, motivating anatomy-aware modeling for improved localization. However, prior anatomy-aware methods typically rely on coarse region proxies or static spatial priors, which may restrict dynamic instance discovery and limit precise localization of small abnormalities. We propose Anatomy-Structured Hierarchical Multiple Instance Learning (ASH-MIL), a framework that introduces parallel anatomy-structured observation branches (cardiac, pulmonary, and agnostic) combined with hierarchical MIL aggregation. Anatomical priors are injected as soft spatial biases into decoder cross-attention, enabling anatomically grounded evidence maps without disease bounding-box supervision. Instance localization is derived directly from MIL-weighted cross-attention maps without bounding box supervision. Experiments on CXR8 and cross-domain MIMIC-CXR held-out sets demonstrate consistent improvements over prior weakly-supervised and anatomy-aware approaches, particularly under stricter localization criteria. Our code is available at https://github.com/jn-kim/ash-mil.

Problem

• weakly-supervised disease detection
• chest X-rays
• anatomy-aware modeling
• instance localization
• multiple instance learning

Innovation

• Anatomy-Structured Hierarchical MIL
• Weakly-Supervised Learning
• Chest X-ray
• Cross-Attention
• Instance Localization