Installation

Stable release (pip)

To install fPCA from PyPI with pip:

$ pip install Neuro-fPCA-fMRI

This installs the latest stable release and the CLI entry points (fpca-main, preprocess-nii-file, nifti-viewer, compare-pcs).

Windows (recommended: conda/mamba)

On Windows we recommend using a conda environment (via Miniconda or Mambaforge) for smooth installation of scientific dependencies:

1. Install Miniconda or Mambaforge

   • Miniconda: https://docs.conda.io/en/latest/miniconda.html

   • Mambaforge: https://github.com/conda-forge/miniforge

2. Create and activate an environment (Python 3.11 suggested)

# Install mamba in base (optional but faster)
conda install -c conda-forge mamba

# Create environment
mamba create -n fpca-env -c conda-forge python=3.11

# Activate environment
mamba activate fpca-env

3. Install core scientific packages from conda-forge (ensures MKL/OpenBLAS are configured correctly)

mamba install -c conda-forge numpy scipy mkl mkl-service pandas matplotlib seaborn scikit-learn scikit-learn-extra imbalanced-learn plotly joblib nilearn nibabel scikit-fda dtaidistance pytorch shap

4. Install the Neuro-fPCA-fMRI package (PyPI) into the same environment

pip install Neuro-fPCA-fMRI

From sources

The sources for Neuro-fPCA-fMRI can be downloaded from the Github repo.

Clone the repository:

$ git clone https://github.com/yanivassf-lab/fmri
$ cd fmri

Install in editable mode (development):

$ conda install -c conda-forge mamba
$ mamba create -n fpca-env -c conda-forge python=3.11
$ mamba activate fpca-env
$ mamba install -c conda-forge numpy scipy mkl mkl-service pandas matplotlib seaborn scikit-learn scikit-learn-extra imbalanced-learn plotly joblib nilearn nibabel scikit-fda dtaidistance pytorch shap
$ pip install -e .

Verify installation

fpca-main --help
preprocess-nii-file --help
fmri-fpca-pipeline --help