[FieldTrip] ERP average Brain Vision is different from ERP average FieldTrip

[Eelke Spaak]

Dear Karen,

It seems likely (as the other responses also indicate) that
BrainVision Analyzer is doing something to the data that FieldTrip is
not; in other words, the FieldTrip ERP is probably more 'pure'.
Therefore, perhaps it might be worth asking around on the BVA mailing
list if the people there know what BVA is doing to the data prior to
computing and displaying the average?

It is easy to check whether FieldTrip is doing something unexpected to
the data, by computing and plotting the average yourself:

erp = mean(cat(3, pp10_l.trial{:}), 3);
chanind = strmatch('AF7', pp10_l.label);
plot(pp10_l.time{1}, erp(chanind,:));

This only works if all trials have identical time axes, but judging
from your script I think they do. It the above steps give a different
plot than the FT functions, something is possibly (/probably) wrong in
the FT code.

Best,
Eelke

On 17 June 2013 15:22, Karen Schuil <schuil at fsw.eur.nl> wrote:
> Dear Jim, Aaron, Roemer and Tineke,
>
> Thanks for your quick response and suggestions! The whole preprocessing
> (including the filters) was done in BVA. After segmentatation of the
> conditions, we exported the data to Fieldtrip. The only steps we did in
> Fieldtrip was averaging, creating and plotting an ERP. Therefore it can't be
> the filters, right? Unless, the averaging step in BVA applies filters as
> well.
>
> The ref in AF7-ref is added by BVA and refers to the channels being linked
> to the mastoids.
>
> We have also tried it with subtracting a baseline, but this unfortunately
> did not help.
>
> Do you have any other suggestions?
>
> Cheers, Karen
>
>
>
>
> On Mon, Jun 17, 2013 at 2:25 PM, Aaron Schurger <aaron.schurger at gmail.com>
> wrote:
>>
>> To me it really looks like BVA is applying a high-pass filter at some
>> stage. When you export the data, the high-pass filter has probably
>> already been applied. It is typical in EEG (though not a good idea in
>> my opinion) to apply a high-pass filter with a cutoff at around 0.05
>> or 0.1 Hz. There should be a setting somewhere in BVA to turn off the
>> high-pass filter. Anyway, that's my guess just from looking at the
>> figure you attached.
>> Cheers,
>> Aaron
>>
>> On Mon, Jun 17, 2013 at 1:56 PM, Karen Schuil <karenschuil at gmail.com>
>> wrote:
>> > Dear Fieldtrip Users,
>> >
>> > I have a BVA-Fieldtrip problem. The average ERPs of Brain Vision
>> > Analyser
>> > are different from the ERPS of FieldTrip. It seems that in Fieldtrip a
>> > slow
>> > negative drift is added and peaks are more/less pronounced than in BVA
>> > (attached is a picture of the two different plots).
>> >
>> > An expert FieldTrip User and I could not find a solution for this
>> > problem. I
>> > hope one of you has a suggestion for this problem.
>> >
>> > The individual trial data was exported from BVA (version 2.02.5859) with
>> > the
>> > following settings:
>> > File extension: .seg
>> > Write header file: yes
>> > Write marker file: yes
>> > Format: BINARY
>> > Orientation: MULTIPLEXED
>> > Line Delimiter: CRLF (PC style)
>> > Binary format: 16-Bit signed integer format
>> > Set resolution manually: no
>> > Individually optimized resolution for each channel: yes
>> > Convert to big-endian order: no
>> > Export all channels: no
>> > Export the following channels:
>> > AF3 AF4 AF7 AF8 AFz C1 C2 C3 C4 C5 C6 CP1 CP2 CP3 CP4 CP5
>> > CP6 CPz Cz F1 F2 F3 F4 F5 F6 F7 F8 FC1 FC2 FC3 FC4 FC5
>> > FC6 FCz Fpz FT7 FT8 Fz Iz O1 O2 Oz P1 P10 P2 P3 P4 P5
>> > P6 P7 P8 P9 PO3 PO4 PO7 PO8 POz Pz T7 T8 TP7 TP8
>> > Created Using Component Version 2.0.2.5827
>> >
>> > We used Matlab version 7.10.0 R2, and fieldtrip version 13-06-2013 (and
>> > tried a version from 2011).
>> >
>> > The scripts we used are read_analyzer_data and timelockanalysis. This is
>> > the
>> > code we used for calling the scripts:
>> > % read data into fieldtrip
>> > cfg = [];
>> >
>> > cfg.inputfile = 'pp10_A';
>> > cfg.triggercode = 'S 20';
>> > cfg.triggertype = 'Stimulus';
>> > cfg.prestim = 1.2;
>> > cfg.poststim = 1.7;
>> >
>> > pp10_l = read_analyzer_data(cfg);
>> >
>> > % check: compute ERP
>> > %
>> > cfg = [];
>> > pp10_ERP = timelockanalysis(cfg, pp10_l);
>> >
>> > %plot ERP
>> >
>> > cfg = [];
>> > cfg.layout = 'elec1010.lay';
>> > cfg.xlim = [-0.15 1.7];
>> > cfg.ylim = [-12.25 12.25];
>> > % cfg.baseline = 'yes';
>> > % cfg.baselinetype = 'absolute';
>> > cfg.showlabels = 'yes';
>> > cfg.interactive = 'yes';
>> >
>> > multiplotER(cfg, pp10_ERP);
>> >
>> >
>> > I hope you can help!
>> >
>> > Kind regards,
>> > Karen
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > fieldtrip mailing list
>> > fieldtrip at donders.ru.nl
>> > http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>>
>>
>>
>> --
>> Aaron Schurger, PhD
>> Post-doctoral researcher
>> INSERM U992 / NeuroSpin
>> CEA - Saclay, France
>> +33-1-69-08-66-47
>> aaron.schurger at gmail.com
>> http://www.unicog.org
>> _______________________________________________
>> fieldtrip mailing list
>> fieldtrip at donders.ru.nl
>> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip
>
>
>
>
> --
> -------------------
> Karen Schuil
> PhD student
>
> Erasmus University Rotterdam
> Institute of Psychology, T 13-09
> Burgemeester Oudlaan 50
> P.O. Box 1738
> 3000 DR Rotterdam
> The Netherlands
> Phone: +31 (0) 10 408 2293
> Email:  schuil at fsw.eur.nl
>
> _______________________________________________
> fieldtrip mailing list
> fieldtrip at donders.ru.nl
> http://mailman.science.ru.nl/mailman/listinfo/fieldtrip