Towards a Virtual Neuroscientist: Autonomous Neuroimaging Analysis via Multi-Agent Collaboration
Researchers have developed NEXUS, a multi-agent framework designed to autonomously analyze neuroimaging data. This system integrates workflow execution with an understanding of scientific objectives, allowing specialist agents to collaboratively build and optimize executable programs. NEXUS aims to overcome the limitations of static workflows by dynamically adapting to runtime observations and employing a hierarchical verification process for quality control. Experiments on ADHD-200 and ADNI datasets showed NEXUS surpassing standard baselines in predictive performance and demonstrating adaptive refinement capabilities. AI
IMPACT Autonomous agents could accelerate biomarker discovery and clinical trial analysis in neuroscience.