Here it is possible to define which channels to analyze, currently all 64 channels, as well as the frequency resolution and time resolution. To avoid edge effect in the time-frequency transform make sure to have a few hundred extra ms. in each end of the epoch in the EEGLAB dataset, currently set the time region to be around –100 ms to 500 ms. The transformation is currently selected to be the complex Morlet wavelet (‘cmor’) with bandwidth parameter (σ = 1) but a short time frequency transform given by the Gabor transform is also implemented. (Note, other continuous wavelets than the complex Morlet can easily be adapted into ERPWAVELAB, see the section ‘how can I define my own wavelets’ in the ERPWAVELAB help under ‘frequently asked questions’). 
Below, the selected wavelet is displayed. Since transforming the data to the wavelet domain in general requires a large amount of RAM, the options to generate smaller datasets only containing specific measures of the wavelet transform is given. If ‘Create All’ is selected the full dataset containing the wavelet transform of all epochs is generated which is needed for ERPCOH analysis and also to calculate bootstrap significance. (Presently, the dataset is reasonably small such that most computers should be able to handle the full time-frequency transformation. If for some reason you run out of memory reduce the resolution of the analysis either by analyzing fewer channels, frequencies or time points or by splitting the dataset into smaller segments in EEGLAB containing fewer epochs prior to generating the datasets in ERPWAVELAB. By loading all the smaller data segments and selecting ‘Use all’ from the GUI (see section on
multiple dataset analysis), the full dataset can still be analyzed.

Click ‘Save to file’ and write the name of the file in which to save the generated ERPWAVELAB dataset. Then click ‘Create’. The EEGLAB dataset is being time-frequency transformed  to generate the ERPWAVELAB dataset - the MATLAB prompt will notify how long this takes.

To ease the process of generating ERPWAVELAB files, the function call that forms the dataset is stored in the EEGLAB variable EEG.history, see EEGLAB for details on saving the dataset history and creating batches. If you type:
EEG.history
In the MATLAB prompt you can see the function call generating the ERPWAVELAB dataset at the end of the command history of the current dataset in EEGLAB.

 

The ERWAVELAB file is stored as a regular .mat file containing the following variables:

Select: File -> Create ERPWAVELAB file -> From EEGLAB (CURRENTSET)
A window opens enabling
you to specify how to perform the time-frequency transform.

 

Creating ERPWAVELAB datasets (continiued)

Text Box: <<PREV
Text Box: NEXT>>

Open ERPWAVELAB if it is not already open. By typing ERPWAVELAB in the directory in which ERPWAVELAB was installed.

Text Box:             ERPWAVELAB

Developed by Morten Mørup

A tOOLbox FOR MULTI-CHANNEL TIME-FREQUENCY ANALYSIS

WT

The wavelet transformed array of channel x frequency x time x epochs

chanlocs

The location of each of the original EEG data channels

tim

The timepoints analyzed corresponding to the indices in the time dimension of WT

Fa

The frequencies analyzed corresponding to the indices in the frequency dimension  of WT

Fs

The sampling rate of the original data

wavetyp

The transform used for  the time-frequency analysis

nepoch

The number of Epochs used in the analysis if the dataset has not been set to generate

the full array containing all epochs