Explainable AI improves accuracy and review in radiology reports
Interleaving anatomical inspection and diagnostic reasoning for explainable radiology report generation
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Radiology report generators can create plausible reports but do not clearly show how they reach their conclusions, making it hard to check or correct mistakes. The authors developed ARIVE, an AI system that explains its diagnostic reasoning by linking visual checks of body parts to clinical findings before writing the report. They trained this system on a large dataset of chest X-rays and showed that it produces reports that experts rate highly and that are easier to trace back to the medical images. Having access to ARIVE’s step-by-step reasoning helps medical reviewers spot errors more effectively.
What this means in practice
- •For hospital imaging teams: Generate radiology reports with traceable diagnostic reasoning to improve accuracy and facilitate expert review and error correction.
- •For medical software developers: Build explainable automated reporting tools that incorporate anatomical inspection and clinical context to produce clinically reliable radiology reports.$Commercial implications: Enables development of commercial AI report generation systems with explainability features attractive to healthcare providers and regulatory bodies.
Authors
Chentao Huang, Runlong Lin, Zhijun Wang, Yunxiang Li, Yuanyuan Sun, Jian Wang, Hongfei Lin, Ling Luo
Abstract
Automated radiology report generation can draft clinically plausible reports, yet the pathway from visual evidence to reported findings remains opaque, leaving the diagnostic basis of these reports unexplained and the generated reports difficult to review or revise. We introduce ARIVE (Anatomy-guided Reasoning with Interleaved Visual Evidence), an explainable report generation framework that interleaves anatomical inspection with diagnostic reasoning prior to final report composition. To support this process, we constructed 159,416 multimodal chain-of-thought (MCoT) trajectories from MIMIC-CXR, wherein visual evidence from inspected anatomical regions informs subsequent clinical interpretation. Trained on these trajectories and further optimized using anatomical and clinical-content rewards, ARIVE achieved superior clinical-efficacy performance compared with strong baselines on MIMIC-CXR and the abnormality-centered MIMIC-ABN. ARIVE's reports were competitive with those of leading models in blinded expert grading, and its MCoT trajectories earned high ratings for diagnostic traceability. Furthermore, a paired review evaluation suggests that access to MCoT can aid in identifying errors within generated reports.