BrainFC

Resting-state fMRI (rs-fMRI) → functional connectivity → brain networks and hypergraphs.

Neuroimaging, connectome visualization and network analysis in one Python package, with a local GUI and complete API. Requires Python 3.11–3.13.

Install and launch

pip install -U brainfc
brainfc serve

Choose 打开真实样例 to try the bundled rest01 example. No files or parameters to supply.

Python API

from brainfc import Config, extract_connectome

# A TSV with a header: rows are time points, columns are ROI signals.
result = extract_connectome(
    "signals.tsv",
    config=Config(detrend=False, standardize=False),
)
matrix = result.connectivity
result.save("results/run-001")

The destination must be new. For volume images, supply an integer-label atlas in the same explicitly named space and confirm that spatial preprocessing is complete. Filtering, confound regression and censoring must match the provenance of the input signals; the example above does not add temporal denoising.

Capabilities

• Read supported NiBabel volume containers, CIFTI time series, paired GIFTI data/labels, and CSV/TSV/TXT/1D/NPY/NPZ/MAT tables (excluding MAT v7.3).
• Extract ROI means; apply confound regression, temporal cleaning and censoring while retaining original frame indices.
• Compute Pearson, Spearman or Ledoit-Wolf partial correlations, with a separate Fisher-z matrix.
• Export arrays, tables, quality records, input fingerprints, figures and an offline interactive report.
• Synchronize selected connections and display thresholds between the 3D viewer and eight anatomical views. Display filtering does not modify the complete signed matrix.

Raw BOLD + T1 and supported DICOM series have an in-process Python preprocessing route with ANTsPy, followed by visual QC and connectivity extraction. No MATLAB or Docker is required. Susceptibility-distortion correction is not implemented. This package does not provide disease diagnosis or cohort-level inference.

Documentation and source

• Quickstart
• Complete Python API
• Python usage guide
• Input formats and processing contract
• Source code and issues

Licensed under Apache-2.0. The 3D viewer is adapted from Hyper-Brain; BrainFC runs independently. Dataset and atlas licenses remain with their original providers.

Network analysis

Choose 进入网络分析 after extraction to explore graphs, native hypergraphs and network metrics. The matrix and ROI mapping transfer automatically.

from brainfc.network import AnalysisConfig

network = result.analyze_network(AnalysisConfig(k=5))

Network guide · Compatibility guide