Skip to content

MRI Data

MRI Data

HyperASHS models are trained on MRI scans acquired with a particular protocol, and they work best on scans that look similar. Before running HyperASHS, check that your data meet the requirements of the model you plan to use:

  • T2w + T1w Model Requirements: T2-weighted MRI (about 0.4 × 0.4 × 2.0 mm³) plus whole-brain T1-weighted MRI, for hippocampal subfields and MTL cortical subregions (PMC_Atlas, HSG_Body_Atlas).
  • T1w-Only Model Requirements: whole-brain T1-weighted MRI only, for anterior/posterior hippocampus and MTL cortical subregions (PMC_T1_Atlas).

You can also see the expected resolution and coverage of each input for any model with:

hrashs desc <model_id>

HyperASHS reads images in NIfTI format (.nii or .nii.gz). Convert DICOM data to NIfTI first, for example with dcm2niix.

Large, public datasets with suitable T2-weighted MTL scans include ADNI3.

Using classic ASHS? See the equivalent page for ASHS Classic.