Find Neuropyhsiological Networks (FiNN)

A Python Toolbox for the analysis of electrophysiological data

The main directory finn contains the toolbox itself, while finn_demo contains a number of demo files describing end-point functionality of the toolbox. finn_tests contains automated tests to perform unit tests in order to evaluate the state of the toolbox.

Installation

FiNN can be installed either

1. manually by downloading a branch of choice
2. via pip when calling pip install finnpy

For more information please click here.

Branches

FiNN offers two branches:

• Stable: Contains only stable releases (passed all unit tests)
• Develop: Contains all latest changes and may be unstable (not unit tested)

Features

Currently implemented features in this toolbox. Full documentation is accessible in doc/_build/index.html. Documentation of the development branch is available here.

• Artifact rejection
  • Identification of bad channels and (optional) restoration of those from neighboring channels
  • Removal of statistical outliers (identified via z-transform)
• Basic
  • Functionality to downsample data
  • Common average re-referencing
• Connectivity
  • Cross frequency coupling
    • Implemented methods are phase lock value, modulation index, mean vector length, and direct modulation index.
  • Same frequency coupling
    • Implemented methods are weighted phase lag index (wPLI), phase slope index, magnitude squared coherence, imaginary coherence, and directional absolute coherence.
• File IO
  • Load data from brain vision recordings.
  • A data manager to save & load (unbalanced) data with a minimal memory footprint.
• Frequency spectrum filters
  • Easy access to scipys butterworth filter
  • An overlap add based implementation of an FIR filter
• Misc.
  • A parallelization loop for easy access to parallel processing. Accustomed to minimal memory footprint and resource consumption
• Statistics
  • Easy within Python access to generalized linear mixed models
• Visualization
  • Plot topographical maps of EEG recordings. Effect sizes may be visualized by color whereas significance may be visualized using dots (black - n.s., half-filled - significant before multiple comparison correction, white - significant after multiple comparison correction)

Requirements

• Python 3.6 or above.
• R and R packages
  • lm4
  • car
  • carData
  • Matrix