MADE-for-ASD: A multi-atlas deep ensemble network for diagnosing Autism Spectrum Disorder

Xuehan Liu, Md. Rakibul Hasan, Tom Gedeon, Md. Zakir Hossain

Abstract: Highlights•Introduces MADE-for-ASD, a deep ensemble network for fMRI-based ASD diagnosis.•Achieves 75.20% accuracy, surpassing prior work by 4.4 percentage points on ABIDE I.•Demonstrates robust sensitivity (82.90%) and specificity (69.70%) across datasets.•Identifies key regions of interest like precuneus and anterior cingulate/ventromedial.•Provides open-access code to promote reproducibility and further research.