arXiv cs.AI· Chentao Huang, Runlong Lin, Zhijun Wang, Yunxiang Li, Yuanyuan Sun, Jian Wang, Hongfei Lin, Ling Luo·· 10 小时前AI 评分38
ARIVE：交错解剖检查与诊断推理的可解释放射报告生成框架
Interleaving anatomical inspection and diagnostic reasoning for explainable radiology report generation
AI 导读
ARIVE 是一个可解释的放射报告生成框架，在最终撰写报告前交错进行解剖检查与诊断推理。研究团队从 MIMIC-CXR 构建了 159,416 条多模态思维链（MCoT）轨迹用于训练，并结合解剖与临床内容奖励优化，在 MIMIC-CXR 和 MIMIC-ABN 上取得优于强基线的临床效能表现。
正文 · 原文
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.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.14018 [cs.CV] |
| (or arXiv:2611.14018v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.14018 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Chentao Huang [view email]
[v1]
Tue, 10 Nov 2026 12:52:23 UTC (26,362 KB)
来源：arXiv cs.AI · arxiv.org