Astra: a generalizable report generation foundation model for 3D computed tomography

2026-05-29

AI Summary

This work addresses the absence of generalizable foundation models for CT report generation, which has hindered consistent diagnostic accuracy and stylistic coherence across multiple anatomical regions and institutions. The study presents the first foundation model tailored for 3D CT report generation, trained on 90,678 chest-abdomen CT–report pairs. By integrating large-scale vision–language pretraining, standardized report styling, reinforcement learning optimization, and multi-organ abnormality modeling, the model effectively mitigates heterogeneity in terminology and reporting style. It achieves an average 44.1% improvement in fine-grained diagnostic metrics across CTRgDB and six external cohorts. Clinical validation demonstrates a 29.6% increase in efficiency for chest report drafting and an 11.3% gain in abdominal report completeness, while also enabling seamless adaptation to diverse downstream AI tasks.

Abstract

CT interpretation requires radiologists to review hundreds of volumetric slices per examination, making reporting time-consuming and highly expertise-dependent. Automated CT report generation offers a promising route to improving clinical efficiency, yet the field still lacks a generalizable CT report generation foundation model that supports multi-region reporting and remains robust across external real-world cohorts. Intrinsic inconsistencies in reporting style and diagnostic terminology across cohorts make naive joint training prone to noisy textual supervision, thereby limiting model generalizability. Here we present Astra, a generalizable CT report generation foundation model trained on 90,678 thoracoabdominal CT-report pairs (CTRgDB) with 353,671 abnormalities spanning eight organ systems. By harmonizing report style and further refining diagnostic consistency via reinforcement learning, Astra achieves style-consistent and diagnostically accurate report generation across diverse anatomical regions and institutions. Evaluating on CTRgDB and six external cohorts, Astra achieves state-of-the-art performance with a 44.1% average improvement in fine-grained diagnostic metrics (P<0.001). In real-world clinical workflows, Astra assistance accelerates chest report drafting by 29.6% and improves abdominal report completeness by 11.3% (P<0.001). Furthermore, Astra also demonstrates broad utility as a foundation for CT AI development, improving downstream diagnostic performance and scaling vision-language pretrain through high-quality report synthesis. Overall, Astra serves as a broadly accessible clinical assistant and a pivotal infrastructure for the next generation of AI-powered healthcare.

Problem

• CT report generation
• generalizable foundation model
• multi-region reporting
• diagnostic consistency
• real-world cohorts

Innovation

• foundation model
• CT report generation
• reinforcement learning
• multi-region generalization
• vision-language pretraining