Difference between revisions of "Export Single Trial Data"

Latest revision as of 15:53, 5 May 2021

Rationale

This article describes how to export single-trial EEG or MEG raw data for further analysis in an external program, e.g. in Matlab.

Steps to follow

Load data in BESA

The data are then displayed in the review window. Now you can either

• use the ERP module to define a condition and reject any trials with artifacts (see the BESA Help section on ERP for more information), or
• directly move to the export of all epochs around a particular trigger.

Export data from BESA

Select "File → Export...". The export dialog opens:

Figure 1 "Export data" dialog

Make sure that the options "Epochs around triggers" and "Original data" are selected. In the section "Target formats", choose the option of choice:

• If you want to create ASCII files for offline analysis, select "ASCII multiplexed" or "ASCII vectorized"
• If you want to analyze data directly in Matlab, select "Send to MATLAB"

For the "Interval..." select a long enough time span as shown below:

Figure 2 "Epoch around trigger" dialog

Then click the "Triggers..." button to select which trigger code or condition will be used.

If you want to export the epochs around a trigger code: Make sure that the trigger code you want is selected - e.g. the trigger code 1 in this example.

Figure 3 "Trigger codes" dialog

If you want to export the epochs around a condition: Check the box "Conditions". Then ensure that the condition of choice is selected (e.g. condition "Rare" in the example in Figure 4).

Figure 4 "Epoch around condition" dialog

After clicking "OK" BESA Research either exports the data, or - if that option was chosen - sends the data to Matlab. In case of Matlab, it opens the Matlab window and creates the structure "besa_channels". From Matlab save the structure as "besa_channels.mat".

Data in MATLAB

The MATLAB structure besa_channels contains several fields like in the following example of a 64-channel EEG file with 39 trials that survived the artifact scan:

              datafile: [1x93 char]
          channeltypes: {1x64 cell}
         channellabels: {1x64 cell}
          channelunits: {1x64 cell}
    channelcoordinates: [64x9 double]
               montage: 'Original'
               filters: 'Filters off'
          samplingrate: 100
        HSPcoordinates: [3x3 double]
              HSPtypes: {'Fiducial'  'Fiducial'  'Fiducial'}
             HSPlabels: {'FidNz'  'FidT9'  'FidT10'}
            headcenter: [-0.0020 8.1247e-004 0.0291]
            headradius: 0.1021
    electrodethickness: 0
            structtype: 'besa_channels'
              datatype: 'Epoched_Data'
                  data: [1x39 struct]

The sub-structure data contains the latencies and the measured amplitudes for each channel and trial:

>> besa_channels.data
 
ans =
 
1x39 struct array with fields:
    amplitudes
    timeoffsetsecs
    latencies
    event