Tim improves radiology reports by tracking disease changes over time
TIM: Temporal Decoupling with Iterative Mutual-Refinement Model for Longitudinal Radiology Report Generation
Computer Vision and Pattern Recognition
Summary
Doctors use images taken at different times to see how diseases change. Current computer programs create reports based on one image at a time and don’t track these changes well. The authors designed TIM, a two-step method that separates what is seen in each image from how diseases evolve over time. It produces better reports by correcting inconsistencies between past and current findings. Tests showed TIM works better than earlier methods on a public medical dataset.
What this means in practice
- •For radiology software engineers: Enhance tools that generate automated radiology reports by incorporating disease progression over time to produce more consistent and accurate clinical summaries.
- •For clinical informatics teams: Integrate improved longitudinal report generation methods to reduce radiologists’ workload and standardize documentation in hospital systems.
- •For medical transcription service providers: Develop automated transcription products that use temporal refinement to improve accuracy of diagnosis descriptions across multiple patient visits.$Commercial implications: The model enables sale of advanced transcription software offering enhanced report accuracy for healthcare providers managing longitudinal patient records.
Authors
Yiheng Dong, Yi Lin, Shilong Huang, Xiyan Yang, Xin Yang
Abstract
Automatic radiology report generation (RRG) aims to translate medical images into diagnostic text, reducing radiologists' workload and standardizing clinical documentation. Nonetheless, existing approaches mainly focus on single-timepoint analysis and fail to capture temporal disease evolution across longitudinal examinations. While recent longitudinal RRG (LRRG) approaches incorporate historical data, they often combine images from different time points within a single representation space, leading to blurred semantics and inconsistent temporal reasoning. In this work, we propose a Temporal Decoupling with Iterative Mutual-Refinement Model (TIM), a two-stage framework that explicitly decouples spatial pathology from temporal progression and iteratively refines reports through mutual feedback. Stage I performs temporal-decoupled representation learning, separating temporal evolution patterns from disease-specific features and generating radiology reports for both prior and current studies. Stage II introduces a mutual report refinement mechanism that identifies diagnostic inconsistencies within prior reports and iteratively rectifies both prior and current reports through error-sensitive feedback. Experiments on the Longitudinal-MIMIC dataset demonstrate that TIM surpasses existing single-image and longitudinal baselines, achieving new state-of-the-art performance across both language and clinical metrics. Code is available at https://github.com/yihengd/TIM.