Skip to content
Time series visualizer and data processing tools for Electrocorticography (ECoG) signals stored in NWB files, for Python.
Python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
docs
ecogvis
media
.gitignore
LICENSE
MANIFEST.in
README.md
make_env.yml
setup.py

README.md

ecogVIS

Timeseries visualizer and data processing tools for Electrocorticography (ECoG) signals stored in NWB files, for Python.

A collaboration with with Chang Lab.

Installation

To clone the repository and set up a conda environment, do:

$ git clone https://github.com/ben-dichter-consulting/ecogVIS
$ conda env create -f ecogVIS/make_env.yml
$ source activate ecog_vis

Alternatively, to install ecogVIS directly in an existing environment:

$ pip install git+https://github.com/ben-dichter-consulting/ecogVIS

After activating the correct environment, ecogVIS can be imported and run from python. If the file does not exist (or if you provide an empty string ''), you'll be prompted to choose a file from a dialog.

from ecogvis.ecogvis import main
import os

fpath = os.path.join('path_to','file.nwb')
main(fpath)

Features

ecogVIS makes it intuitive and simple to viualize and process ECoG signals. It currently features:

Navigation Seamless visual navigation through long signals from large arrays of electrodes, by mouse-dragging visualization window, control buttons, value fields and keyboard keys.

Annotations Add, delete, save and load annotations for meaningful comments anywhere in the visualization.

Intervals Add, delete, save, load and create custom intervals types to mark specific points in time, with simple click-drag-release mouse movements.

Bad Channels Mark and un-mark bad channels. Choices are saved in the electrodes group of the current NWB file.

Signal preprocessing Preprocessing of raw voltage signals, including user-defined Downsampling, CAR and Notch filtering. The resulting processed signals are stored as an LFP object, in the processing group of the current NWB file.

Events detection Automatic detection of events in audio recordings for Consonant-Vowel tasks. The audio data should be stored in the NWB file in the following way:
  • Speaker audio - As a TimeSeries object, named 'Speaker CV', in the stimulus group.
  • Microphone audio - As a TimeSeries object, named 'Microphone CV', in the acquisition group.
The resulting detected intervals, 'TimeIntervals_mic' and 'TimeIntervals_speaker', are saved as TimeIntervals objects in the intervals group of the current NWB file and can be used later for ERP analysis. A preview allows for testing of the detection parameters before running it for the whole duration of the audio signals.

High Gamma Estimation of high gamma analytic amplitude, with the average of user-defined specific bands. The results are saved as a TimeSeries object, named 'high_gamma', in the processing group of the current or of a new NWB file.

Event-Related Potentials Grid visualization of high gamma ERP calculated in reference to:
  • Stimulus (speaker) or response (microphone) time intervals
  • Onset or offset points
The grid items are coloured to mark specific cortical areas and can be rotated to correspond anatomically to them. Emphasis can be given to specific areas of interest and double-clicking an item allows for fast inspection of the single electrode's ERP in detail.

Periodograms Grid visualization of Periodograms:
  • Raw and preprocessed data
  • FFT and Welch methods
  • Individual devices
The grid items are coloured to mark specific cortical areas and can be rotated to correspond anatomically to them. Emphasis can be given to specific areas of interest and double-clicking an item allows for fast inspection of the single electrode's Periodogram in detail.

Spectral Decomposition Analytic signal amplitude estimation by Hilbert transform of user-defined frequency bands.

WARNING: This function will manipulate an array of size (nSamples, nChannels, nBands), which might be in the order of gigabytes and demand a large memory to operate and is likely to produce a large file.

Plus

  • Select electrodes from specific brain areas
  • Easy moving between block files for the same subject
You can’t perform that action at this time.