Open-weight brain MRI, head CT & fMRI foundation models on one PACS bench.
Select a 3D neuroimaging foundation model, feed it NIfTI volumes (.nii / .nii.gz), DICOM series, or slice stacks, and inspect the output its training was built for: 3D patient embeddings, BrainAge & morphometry, urgent MRI/CT diagnostic triage, structured neuroradiology reports, tri-planar parcellation & lesion masks, cross-sequence synthesis, or 4D fMRI functional networks — with verified links to every paper. Inference runs on your own workstation or GPU server; this static page never touches patient scans.
Run a neuroimaging model
Jobs queue on your local/remote GPU server (or run instantaneously in browser Demo mode with an analytical 3D brain phantom).
Results
Studies & jobs
Papers, BibTeX & neuroimaging benchmarks
Verified journal DOIs, arXiv/medRxiv preprints, and one-click BibTeX citations for all 28 open-weight brain radiology foundation models, alongside standard neuroimaging challenges (BraTS, FOMO-MRI, ATLAS, RSNA ICH, ADNI/OpenNeuro, UK Biobank) and ecosystem references.
Connect a neuroimaging backend
GitHub Pages hosts this static workstation UI. Real 3D NIfTI / DICOM inference runs inside the lightweight FastAPI server in backend/ on your local workstation, hospital research server, or cloud GPU.
Server status
Connect to inspect CUDA GPUs, NiBabel/MONAI packages, and cached model weights.