Radcarve improves radiology report accuracy by matching anatomy and text
Carving Radiology Reports at Their Evidential Joints: Anatomical Region Sets as Generation Units
Computation and LanguageComputer Vision and Pattern Recognition
Summary
Radiology reports describe what doctors see in medical images, but organizing them into smaller, clear pieces that link to specific body parts is hard. The authors created RadCarve, which breaks down reports based on precise anatomical regions and generates each piece of text from the related image areas. This method helps create more accurate and detailed reports, better matching the text to the visuals of the body. RadCarve also uses fewer computing resources while outperforming existing tools on large datasets.
What this means in practice
- •For radiology software developers: generate more accurate and anatomically precise report sections directly tied to image features for improved clinical text production.
- •For medical imaging platform teams: enhance automated report generation pipelines to better bind textual descriptions with specific anatomical regions in images, improving report detail quality.
Authors
Jinwei Yang, Junhao Chen, Jianwei Zhao, Zhonghan Xu, Hong Cheng, Ling Wang
Abstract
Organizing radiology reports into independently generated units offers a way to generate fine-grained clinical content and bind it to visual evidence. However, existing methods typically adopt generation-unit structures not directly specified in natural reports or their annotations, introducing additional modeling compromises. We propose RadCarve, which instead defines generation units directly from the anatomical region sets in existing statement-region correspondences, unifying unit formation with visual evidence selection. RadCarve's planner uses composition-specialized query banks to jointly predict unit existence and region composition from the image, iteratively refining both with regional evidence aggregated according to current membership predictions. The generator independently produces each unit's text conditioned on the joint visual evidence of its predicted region set. Compared with strong baselines, RadCarve improves CXRFEScore by 9.4% in unit generation given region sets and achieves 2.4 times the end-to-end binding F1. It also achieves leading report-generation results on Chest ImaGenome and PadChest-GR, using fewer than one-sixth as many parameters as large-scale baselines.