home - installation - data preprocessing and editing - data review and visualization  - statistics and extended data analysis - user interface reference

Transposing, filtering, and rereferencing

The second analysis step involves rereferencing, transposing and filtering of the raw data file. Physiological data is nearly always saved to disk in a channel-by-samplepoint order by the measuring device. For filtering and other processing steps however, it is much more useful to have the data in a samplepoint-by-channel order, that is all values of on channel first, then all values of the next channel and so on, therefore the data is always reordered (or transposed) first and - in order to keep the following steps independant of the original dataformat -  transformed to the EGI data format. The resulting files have the file name extension *.TAW .

While rearranging the data, EMEGS can also rereference the channels. BDF files for instance always need to be rereferenced first, to get a better signal quality. But this step can also be used to transform the data from a single mastoid reference to an average mastoids reference.

To do this, you need to adjust your sensor configuration file by adding reference weights in the "RefWeight" column. 

The case shown below is for rereferencing a BDF file to an average-mastoids-reference (from the Biosemi CMS recording reference). Note the "RefWeight" column, which contains positive non-zero weights for the mastoid sensors (A1 and B1) and zero for all others.

If all sensors have zero weights, no rereferencing is performed.

Type   Name    InRawFile     RefWeight      Occular       Theta        Phi         Rho
EEG    Fp1     Yes           0              No            -1.6057      -1.25664    0.09
EEG    AF7     Yes           0              No            -1.6057      -0.942478   0.09
EEG    AF3     Yes           0              No            -1.29154     -1.13446    0.09
EEG    F1      Yes           0              No            -0.872665    -1.18682    0.09

...    ...     ...           ...            ...           ...          ...         ...

EEG    LA1     Yes           0              HEOG_L        -1.96175     -0.842842   0.09
EEG    LA2     Yes           0   
          LVEOG_B       2.17302      1.9583      0.09
EEG    LA3     Yes   
       0              LVEOG_T       1.77549      1.8746      0.09
EEG    RA1     Yes   
       0              HEOG_R        1.96175      0.842842    0.09
EEG    RA2     Yes   
       0              RVEOG_B       2.17302      1.1833      0.09
EEG    RA3     Yes   
       0              RVEOG_T       1.77549      1.26698     0.09
EEG    A1      Yes   
       0.5            No            -2.00713     0.1         0.09
EEG    B1      Yes   
       0.5            No            2.00713      -0.1        0.09
STATUS Status  Yes   
       0              No            NAN          NAN         NAN

PLEASE NOTE: Do not use this mechanism to transform the data to an average reference. EMEGS can do this automatically in a much more convenient way, excluding corrupted sensors from the average!!!

Filters (lowpass, highpass and stopband) are activated and configured on the filter dialog, that is shown before transposing, rereferencing and filtering is started. Activate the filters you need, push the configure-buttons to adjust the cutoff-frequencies and filter orders, and press OK.



After the first activation of a filter, a configuration dialog is shown automatically (which can be reopened later on using the "configure" button) to configure filter details:




After filtering with a 40Hz lowpass filter and rereferencing to average-mastoids the datafile shown on the previous page, looks like this:



In your data folder, some new files have appeared: an empty (dummy) file for the rereferenced and transposed data file, the filtered, rereferenced and transposed data file, and an individual sensor configuration file. The dummy file is needed to allow EMEGS to skip the transposing later on, if the "skip-step-if-file-exists"-option is activated. The individual sensor configuration contains the possibly reduced set of channels in the transposed data file.

The filtered file is named according to your filter selection, e.g. fl40 denotes a 40Hz lowpass filter, fh1 denotes a 1Hz highpass filter etc.