Med3D-R1: Incentivizing Clinical Reasoning in 3D Medical Vision-Language Models for Abnormality Diagnosis

Haoran Lai, Zihang Jiang, Kun Zhang, Qingsong Yao, Rongsheng Wang, Zhiyang He, Xiaodong Tao, Wei Wei, S. Kevin Zhou · 2026

Read Med3D-R1: Incentivizing Clinical Reasoning in 3D Medical Vision-Language Models for Abnormality Diagnosis by Haoran Lai, Zihang Jiang, Kun Zhang, Qingsong Yao, Rongsheng Wang, Zhiyang He, Xiaodong Tao, Wei Wei, S. Kevin Zhou (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-01200.txt.

Description

Haoran Lai and co-authors build Med3D-R1, a train plan that spurs clinical thought in 3D medical vision-word models for flaw spotting. Chunky 3D scans, report-pattern overfit, and weak reward design block strong thought in such models. A first tune stage bridges 3D traits to word vectors and re-weights flaw tokens to cut report slant, then a reward stage reshapes the coherence prize to spur step-wise diagnosis. On CT-RATE and RAD-ChestCT choice tests, the model topped prior marks on both sets.

Catalog record

CREATORHaoran Lai, Zihang Jiang, Kun Zhang, Qingsong Yao, Rongsheng Wang, Zhiyang He, Xiaodong Tao, Wei Wei, S. Kevin Zhou
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE36649
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.01200

Browse the full library · About the Tower of Babel