BandRatios

BandRatios project repository: exploring frequency band ratio measures in electrophysiological data.

Overview

Band ratios are a common measure in neuroscience, and are commonly used across cognitive and clinical investigations. In band ratio measures, the ratio of power between two frequency bands are examined as a feature of interest and potential biomarker in M/EEG, ECoG, and LFP data analyses.

In this project, we explore the properties of band ratio measures, and how they relate to other spectral features. Specifically, we examine if band ratio measures are specific to periodic power, and to what degree they reflect other periodic and aperiodic spectral features.

This project was completed in the VoytekLab at UC San Diego, by Thomas Donoghue, Julio Dominguez, and Bradley Voytek.

Project Guide

You can step through all the analyses and results in this project by stepping through the notebooks.

If you want to re-run the project, there are some parts that are done by stand-alone scripts, which are available in the scripts folder. These scripts are also described in the notebooks.

Reference

This project is described in the following paper:

Donoghue T, Dominguez J & Voytek B (2020). Electrophysiological Frequency Band Ratio Measures 
Conflate Periodic and Aperiodic Neural Activity. eNeuro, 7(6) DOI: 10.1523/ENEURO.0192-20.2020

Direct Link: https://doi.org/10.1523/ENEURO.0192-20.2020

Requirements

This project was written in Python 3 and requires Python >= 3.7 to run.

If you want to re-run this project, you will need some external dependencies.

Dependencies include 3rd party scientific Python packages: - numpy - pandas - scipy - matplotlib - seaborn

You can get and manage these dependencies using the Anaconda distribution, which we recommend.

In addition, this project requires the following dependencies:

• mne >= 0.18.2
• fooof >= 1.0.0
• lisc >= 0.1.0

You can install the extra required dependencies by running:

pip install mne, fooof, lisc

Repository Layout

This project repository is set up in the following way:

• bratios/ is a custom module containing code used for analyses and visualizations
• data/ includes data for the project, include simulated data, processed EEG data, and output files
• figures/ contains figures for the project, which are created in notebooks/ and scripts/
• notebooks/ is a collection of Jupyter notebooks that step through the project and create the figures
• scripts/ contains stand alone scripts that run parts of the project

Data

This project uses simulated data, literature data, and electroencephalography (EEG) data.

The simulations are all done using the FOOOF tool and associated simulation framework. Code for generating these simulations is included in this repositry, and all simulated data reported upon is available in the data/ folder.

The literature data was collected with LISC, a tool for collecting and analyzing literature data. Code to re-run the literature data collection is available in the notebooks/. All collected literature data used in this project is available in the data/ folder.

The EEG data is open-access data from the 'Multimodal Resource for Studying Information Processing in the Developing Brain' or MIPDB dataset. This dataset was collected and released by the ChildMind Institute. Raw data can be downloaded through their data portal. The processed power spectra, upon which we operate, and the calculated output measures for this project are collected and available in the data/ folder.