Reproduce “A frontier machine intelligence for human neuroimaging” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Chenhui Zhao, Akshay Rao, Akhil Kondepudi, Yiwei Lyu, Xinhai Hou, Rikhil Seshadri, Cheng Jiang, Soumyanil Banerjee and 14 more

Neuro-VL is a conversational system for interpreting neuroimaging studies, combining a hierarchical visual encoder with a language model. It was trained on nearly one million clinical studies paired with radiology reports. On 17 task-group combinations, it exceeded GPT-5.4 by 15.9 percentage points on an equal-weighted average. These retrospective results do not establish clinical benefit or safety; prospective evaluation remains necessary.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.