New AI Reads Brain Scans to Flag Autism, Depression and Alzheimer's
One brain-scan AI could help doctors spot three hard-to-detect disorders earlier.
Researchers built HyperAMS-Net, a deep learning framework that classifies brain disorders using neuroimaging representations derived from resting-state functional MRI or structural MRI. It learns data-driven weights over multiple receptive fields to capture complementary patterns at different scales — like examining a photo both up close and from far away — and combines hypergraph attention, spatial-channel attention, and adaptive feature fusion. Evaluated on three benchmark datasets spanning distinct brain disorders — ABIDE for autism spectrum disorder, REST-meta-MDD for major depressive disorder, and ADNI for Alzheimer's disease — using 5-fold stratified cross-validation, it attained the highest accuracy and AUC among the compared methods. Ablation studies showed each proposed component contributes, with the largest performance degradation when hypergraph attention is removed. The work was accepted at the 17th International Workshop on Machine Learning in Medical Imaging (MLMI 2026), held in conjunction with MICCAI 2026.
- The AI learned from three public brain-scan datasets covering autism, major depression and Alzheimer's, and outperformed rival methods on each.
- Its strongest component notices groups of brain regions that activate together — not just pairs — which is closer to how real brains work.
- Nothing is available to patients yet: it's peer-reviewed research, not an approved medical device, and MRI scans alone can't diagnose anyone.
Why It Matters
Could one day cut the years-long wait for a brain-disorder diagnosis — but today it's only lab research.