NeuroMosaic: Anatomically Grounded Multimodal Large Language Modeling for Molecularly Aware Glioma Reasoning from 3D MRI and Clinical Narratives

Yantong Liu, Zheyu Zhang, Runpeng Liu +3

Multimodal medical large language models remain structurally weak for neuro-oncology because volumetric evidence is compressed into generic visual tokens and diagnostic conclusions…