CheXRelNet: An Anatomy-Aware Model for Tracking Longitudinal Relationships between Chest X-Rays
This work addresses a specific problem in medical imaging for healthcare professionals, focusing on monitoring disease progression in chest X-rays, and is incremental as it builds on existing deep learning methods for chest radiograph interpretation.
The paper tackles the problem of tracking longitudinal pathology changes between sequential chest X-rays, which is challenging due to anatomical motion and image registration issues, and reports increased downstream performance on the Chest ImaGenome dataset compared to baselines.
Despite the progress in utilizing deep learning to automate chest radiograph interpretation and disease diagnosis tasks, change between sequential Chest X-rays (CXRs) has received limited attention. Monitoring the progression of pathologies that are visualized through chest imaging poses several challenges in anatomical motion estimation and image registration, i.e., spatially aligning the two images and modeling temporal dynamics in change detection. In this work, we propose CheXRelNet, a neural model that can track longitudinal pathology change relations between two CXRs. CheXRelNet incorporates local and global visual features, utilizes inter-image and intra-image anatomical information, and learns dependencies between anatomical region attributes, to accurately predict disease change for a pair of CXRs. Experimental results on the Chest ImaGenome dataset show increased downstream performance compared to baselines. Code is available at https://github.com/PLAN-Lab/ChexRelNet