Anatomy aware method improves chest X ray disease localization
Anatomy-Structured Hierarchical MIL for Weakly-Supervised Thoracic Disease Detection in Chest X-rays
Computer Vision and Pattern Recognition
Summary
Detecting diseases in chest X-rays is hard because many organs overlap and some illness signs are very small. The authors created a method that looks separately at heart, lungs, and general areas to better find where disease might be. Their approach uses soft hints about anatomy to guide the computer's attention without needing detailed maps of disease locations. Tests on different chest X-ray collections showed their method worked better than earlier ones for spotting disease areas. This could help in identifying subtle signs of illness more precisely using only simple labels.
What this means in practice
- •For hospital data teams: Improve automated detection of specific disease areas in chest X-rays using only image-level diagnosis labels.
- •For medical imaging developers: Build X-ray diagnostic tools that provide better localization of diseases within anatomically relevant regions without extra annotation costs.
Authors
Jeongin Kim, Sohyun Ahn, Seo Young Kang, Jaeyi Sung, Soomin Kim, Sungho Cho, Rena Lee, Kwanchang Kim, Junhyug Noh
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.