Getting started¶

This guide walks through installing parcellate and running your first parcellation.

Installation¶

pip install parcellate

Or install from a local checkout:

git clone https://github.com/GalKepler/parcellate.git
cd parcellate
pip install -e .

Verifying your environment¶

parcellate depends on Nibabel, NumPy, and pandas. To confirm your installation:

>>> from parcellate import VolumetricParcellator

If that import succeeds, you're ready to go.

Quick start¶

The snippet below demonstrates the essential steps: load an atlas, connect a lookup table, and compute parcel-wise statistics.

import nibabel as nib
import pandas as pd
from parcellate import VolumetricParcellator

# Create the parcellator
parcellator = VolumetricParcellator(
    atlas_img="atlas.nii.gz",
    lut="atlas_lut.tsv",  # TSV with "index" and "label" columns
)

# Fit and transform a scalar map
parcellator.fit("subject_T1w.nii.gz")
regional_stats = parcellator.transform("subject_T1w.nii.gz")
print(regional_stats.head())

The output is a pandas.DataFrame with one row per atlas region and one column per statistic. By default all 45 built-in statistics are computed. Use stat_tier="core" or stat_tier="extended" to compute fewer columns.

Next steps¶

Guide	Description
Usage guide	Atlases, masks, resampling, probabilistic atlases, custom statistics
Metrics reference	All 45 statistics organized by tier
API reference	Full API documentation