GRCD: Grounded Region Change Detection for Multi-Finding Chest X-Ray Pairs

2026-07-02

AI Summary

Existing methods for generating comparative reports from chest X-rays struggle to simultaneously achieve multi-lesion localization, spatial specificity, and modeling of temporal changes. This work addresses these limitations by constructing a curated longitudinal chest X-ray dataset and introducing a novel framework that incorporates region-guided change tokens and a dual-path fusion mechanism—combining upfront spatial tokens with gated cross-attention—to inject anatomical region-specific temporal dynamics into the language model. The proposed approach is the first to enable localized report generation capturing mixed-direction changes across multiple lesions, while effectively mitigating label noise. Experimental results demonstrate that the model significantly outperforms current baselines on a multi-lesion test set, achieving consistent improvements in text quality, clinical accuracy, and change detection performance.

Abstract

Radiologists routinely compare current and prior chest X-rays to track disease progression, producing follow-up reports that describe multiple findings, each localised to an anatomical region and annotated with a temporal change status. Existing automated methods either generate reports from a single image without modelling temporal context, or incorporate temporal information but do not ground their outputs spatially. The few approaches that combine temporal reasoning with spatial grounding are restricted to single-finding descriptions, leaving multi-finding reports with mixed change directions unaddressed. We present GRCD, a framework for grounded report generation from chest X-ray pairs in the multi-finding setting. We first construct a rigorously cleaned dataset of temporal chest X-ray pairs by identifying and correcting two systematic labelling errors in the source annotations. We then introduce a Region-Guided Change Token module that encodes per-region temporal change across anatomical structures and injects this signal into a language model through a dual-pathway strategy combining prepended spatial tokens with gated cross-attention. On a multi-finding test set, GRCD outperforms existing baselines on text generation and clinical accuracy metrics, with gains in change detection. Ablation studies confirm that the dual-pathway design outperforms either integration strategy in isolation on text and clinical metrics, and that region-level change encoding is necessary for multi-finding generation. Code is available at https://github.com/UTSA-VIRLab/GRCD

Problem

• change detection
• chest X-ray
• multi-finding
• temporal reasoning
• spatial grounding

Innovation

• Grounded Region Change Detection
• Multi-Finding Chest X-Ray
• Temporal Reasoning
• Spatial Grounding
• Dual-Pathway Integration