Reproduce “MM-NeuroOnco: A Multimodal Benchmark and Instruction Dataset for MRI-Based Brain Tumor Diagnosis” 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
Feng Guo, Jiaxiang Liu, Yang Li, Qianqian Shi, Mingkun Xu
The authors introduce MM-NeuroOnco, a brain-tumor MRI dataset and benchmark that organize image features into questions about diagnosis and tumor attributes. They derive attributes from segmentation masks and extracted radiological signs, then fine-tune a Qwen3-VL-8B model with chain-of-thought supervision. It reaches 51.4% diagnosis accuracy, versus 41.9% for Gemini-3-Flash; results concern a chiefly single-slice benchmark, not clinical performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.