HARFI

High angular resolution functional imaging for fMRI

View repository ↗

MIT License

HARFI (High Angular Resolution Functional Imaging) extends the idea of high angular resolution diffusion imaging to functional MRI, estimating orientation-dependent functional correlations. The processing pipeline (the MATLAB source lives in HARFI_Scripts) is provided as a pre-built Singularity container.

Usage

The pre-built Singularity container is available on Zenodo (DOI: 10.5281/zenodo.15376547). Provide the fMRI image, brain mask, and reference image in an input directory, then:

singularity run \
    --bind [INPUT_DIR]:/input,[OUTPUT_DIR]:/output \
    [PATH_TO_HARFI_SIF_FILE] \
    [R] [DISCRETE] [TR] [N] \
    [FMRI_IMAGE_NAME] [BRAIN_MASK_NAME] [REFERENCE_IMAGE_NAME] [OUTPUT_NAME]

where R is the radius of integration (voxels), DISCRETE the discrete steps in integration, TR the repetition time (seconds), and N the number of volumes to remove from the beginning. The outputs are OUTPUT_NAME.mat (parameters and results), OUTPUT_NAME.nii.gz (correlation maps), and OUTPUT_NAME_SH_even4.nii.gz (FODs).

Zhiyuan Li, Kurt G. Schilling, and Bennett A. Landman. “Robust containerization of the High Angular Resolution Functional Imaging (HARFI) pipeline” Neuroinformatics, 2026.