12
Model Architectures
NeuroSTORM provides tools for fMRI preprocessing, self-supervised pretraining, supervised learning, and downstream analysis. Explore 12 models through a shared supervised training workflow and 15 datasets, including ABIDE I and ABIDE II.
Model Architectures
Task Categories
Input Formats
Datasets
End-to-end preprocessing for volumetric fMRI data with skull stripping, registration, and brain extraction.
Extract and analyze region-of-interest (ROI) time series with Harvard-Oxford atlases and custom parcellations.
Support for SwiFT contrastive learning and MAE self-supervised methods on large-scale datasets.
Twelve supervised backbones through a common training workflow, including Brain-JEPA, BAnD, BrainLM, and TFF.
Fifteen datasets, including HCP cohorts, UK Biobank, ABCD, ABIDE I and II, and clinical and task-fMRI datasets.
Age/gender prediction, phenotype prediction, disease diagnosis, fMRI re-identification, and task state classification.
Browse our models, datasets, tasks, and documentation to get started with NeuroSTORM.