Spectral Mass

Spectral Mass is an EEGLAB Extension (formerly called plugin) used to compute the probability that two EEG datasets are similar in terms of both power (ERSP in EEGLAB terms) and coherence (ITC). It is open source (GPL), hence comments and additions to the code are welcome. The details of this method are discussed in an upcoming paper.

Download Version 0.1

This is the first release of the extension and has been used to test the effectiveness of the extension. This release has been tested only on one dataset and might contain various bugs. As will be discussed in the article, this extension calculates a single p-value, representing the combined probability of a difference in EEG activity between two conditions. 

Development Versions

If you want to contribute or have access to the latest code you can visit our github repository: https://github.com/HIIT/spectralmass

Installing the plugin

The plugin can be installed by uncompressing its zip file in your eeglab13_x/plugins folder. A spectralmass0.1 folder should be created within, containing all extension files.

Step-by-step guide

Using the command "which eeglab" should show the folder in which eeglab is currently installed: 

>> which eeglab
/Users/marco/Documents/MATLAB/eeglab13_2_2b/eeglab.m

This means that the plugins folder is located, in my case, it the "eeglab13_2_2b" folder. The zip file should then be uncompressed in:

/Users/marco/Documents/MATLAB/eeglab13_2_2b/plugins/

This is enough to allow eeglab to load the plugin on startup. The correct behaviour can be verified by looking the eeglab output on launch:

>> eeglab
[...] (other plugins being loaded)
EEGLAB: adding "spectralmass" v0.1 (see >> help eegplugin_spectralmass)

This indicates that the plugin was loaded correctly and is ready for use.

Using the plugin

This version of the extension requires two datasets to be loaded in memory. This can be done by loading existing eeglab datasets as follows:

It is also possible to load datasets in other formats. Spectral Mass requires our datasets to be already event-locked (number of epochs > 1) and already loaded in memory (they appear in the datasets menu). Instructions on how to segment data are provided in the EEGLAB tutorial.

The Spectral Mass plugin can be now started from the "Tools" menu.

This will load the main window. The settings depend on the lenght of your time locking window (this window size was chosen to match the event-locking in Countering Countermeasures: Detecting Identity Lies by Detecting Conscious Breakthrough). The upper and lower frequencies can also be adjusted, depending on your experimental setup (these settings were used to analyse RSVP data presented at a 100ms SOA, hence our upper frequency bound stops at 8 Hz).

Pressing "ok" will start the procedure. This will load show the time-frequency difference between the two conditions and start the randomisation process. This might take long, depending on the number of randomisation samples you chose in the previous step.

After the procedure is complete, you should see the calculated p-value in Matlab's output:

Spectral Mass p-value (difference between LIEN2F-S1_Probe-Long.set and LIEN2F-S1_Catch1-Long.set): 0.598
>>