Research questionCan compact pretrained brain MRI models transfer across Alzheimer’s tasks and cohorts without task-specific retraining?Limited labeled neuroimaging data makes task-specific deep learning difficult. It remains uncertain whether features learned for one brain MRI task generalize to different Alzheimer’s-related tasks and cohorts. Latest papersRecent research connected to this question, newest first.A Generalizable Feature Extractor for Alzheimer's-Related Brain MRI TasksThe evidence concerns a 3D CNN pretrained for brain-age prediction, with its 7.18 million weights frozen and approximately 1% additional parameters trained through LoRA. Experiments cover dementia classification on ADNI, unchanged transfer to OASIS-3, MCI progression prediction using age and a cognitive score, amyloid-positivity prediction, and estimation of normalized hippocampal and white-matter-hypointensity volumes from T1-weighted MRI. Reported results include AUCs of 0.964 on held-out ADNI folds, 0.871 on OASIS-3, 0.828 for MCI progression, 0.804 for amyloid positivity, and R² values of 0.80 and 0.91 for volume estimation.research paper · Sep 4, 2026