pydfc

pydfc implements multiple analytical methods to assess and compare dynamic functional connectivity (dFC) in human fMRI data to evaluate how methodological choices affect dFC estimates.

Key Features:

• Python implementation: Provided as a Python toolbox for computational analysis of dFC.
• Multiple dFC methodologies: Implements seven distinct dFC assessment methodologies as stated in the source study.
• Multi-analysis approach: Enables combined application of multiple methods to capture a broader spectrum of dFC variations.
• Similarity metrics: Computes pairwise similarity metrics across overall, temporal, spatial, and inter-subject dimensions.
• Method clustering: Supports grouping of methods into three clusters that reflect distinct assumptions and inter-group variability.
• Dataset application: Applied to fMRI data from 395 subjects from the Human Connectome Project.
• Physiological confound assessment: Facilitates distinction between neural-driven connectivity changes and physiological confounds.
• Validation framework: Enables assessment and validation of dFC methods under known ground truths.

Scientific Applications:

• Comparative method evaluation: Quantifies how analytical choices influence dFC estimates across methods.
• Exploration of brain dynamics: Characterizes temporal, spatial, and inter-subject variability of functional connectivity in the human brain.
• Biomarker investigation: Assesses reproducibility and variability of dFC patterns relevant to potential biomarkers.
• Confound separation: Differentiates neural-driven connectivity changes from physiological artifacts in fMRI data.
• Method validation: Validates dFC assessment approaches against known ground truths and comparative analyses.

Methodology:

Implements seven dFC assessment methods, computes pairwise similarity metrics across overall, temporal, spatial, and inter-subject dimensions, groups methods into three clusters, and applies analyses to fMRI data from 395 Human Connectome Project subjects.