[FieldTrip] automatic artifact rejection based on voltage threshold
Ivaylo Iotchev
ivaylo.iotchev at gmail.com
Thu May 25 18:30:05 CEST 2023
So this is the loop:
data4.label %at this point of the script the data we
are about to clean is called data4, cleaning happened after other
manipulations (e.g. filter)
chs_remaining=[5:8]; % which channels to use for the cleaning, in our
case this was 5th to 8th
excluded = []; %create matrix where excluded trials
are collected
for i = 1:size(data4.trial,2) %we specify here that we are going
through the trial indexes, i is defined as the number of trials
dat=data4.trial{i}(chs_remaining,:); %we select the channels we
want to clean from trial nr. i
if any(any((dat>abs_thrs))) || any(any((dat<-abs_thrs))) % abs_thrs
- absolute threshold, set at the begining of the script, this line applies
thresholding
excluded=[excluded i]; %we add a trial to excluded if above
condition is met
end
if ~ismember(i,excluded) %the next lines select trials which
survived the absolute threshold cleaning
maxn=size(dat,2)-100;
for n=1:maxn
if
any(max(dat(:,n:(n+100)),[],2)+abs(min(dat(:,n:(n+100)),[],2)) > rel_thrs) %
rel_thrs - relative threshold, set at the begining of the script
excluded=[excluded i];
end
end
end
end
cfg=[];
cfg.trials = setdiff(1:size(data4.trial,2),excluded);
data5=ft_redefinetrial(cfg,data4); %with these three lines we can
finally integrate our cleaning with the FieldTripian ft_redefinetrial :)
I hope this helps. I can trace this work back to my colleagues Marianna
Boros and Boglárka Morvai, but it is possible that these parts have an even
older origin. Some additional comments were now added by me to support
better understanding, to the extent to which I could do so :)
Best wishes, and success!
(Dr.) Ivaylo (Ivo) Iotchev
Am Do., 25. Mai 2023 um 16:11 Uhr schrieb eleonora parrotta <
eleonora_p at hotmail.it>:
> That would be awesome, Dr. Ivaylo!
> I've been trying to implement it for weeks!
> Thank you so much for any help
>
> Eleonora
>
>
> *Eleonora Parrotta, Doctoral Researcher in Social Neuroscience *
> School of Psychology | University of Aberdeen | Aberdeen | AB24 3FX
> e.parrotta.20 at abdn.ac.uk
>
> ------------------------------
> *Da:* Ivaylo Iotchev <ivaylo.iotchev at gmail.com>
> *Inviato:* giovedì 25 maggio 2023 14:50
> *A:* FieldTrip discussion list <fieldtrip at science.ru.nl>
> *Cc:* eleonora parrotta <eleonora_p at hotmail.it>
> *Oggetto:* Re: [FieldTrip] automatic artifact rejection based on voltage
> threshold
>
> Hey Eleonora,
>
> I might be able to send you some lines tomorrow which implement
> thresholding, these lines do not make direct use of FieldTrip I think, but
> they are MATLAB intern and thus can be coordinated within a script that
> uses Fieldtrip functions, too (basically that's what we are doing at the
> moment).
>
> Best wishes,
>
> (Dr.) Ivaylo (Ivo) Iotchev
>
> Am Do., 25. Mai 2023 um 15:45 Uhr schrieb eleonora parrotta via fieldtrip <
> fieldtrip at science.ru.nl>:
>
> Dear Fieldtrippers,
> I am trying to implement an automatic artifact rejection on epoched data.
> The aim of this process would be to compare the same dataset, analyzed
> both in fieldtrip and eeglab. To do so, I would need to efficiently
> implement the same EEGLAB parameters used for the artifact rejection in
> fieldtrip.
> I have two options
>
> 1. eliminate the trials based on a peak-to-peak artifact rejection,
> which detects and remove trials in which the EEG signal exceeds of a
> threshold of 100 microvolt
>
>
> 1. eliminate the trials based on simple threshold voltage, that is,
> eliminate trials in which the voltage exceeds of +- 50 microvolts.
>
> Both options would be fine as both are implemented in eeglab, but I do not
> manage to implement neither of the two into fieldtrip.
> I've tried to use both ft_artifact_threshold and ft_rejectartifact, even
> by combining the two.
>
> ft_rejectartifact is not able not detect artifacts in the data, the output
> is always :
>
> %artifact rejection
> cfg = [];
> cfg.artfctdef.channel = [1:28 31:65];
> cfg.artfctdef.threshold.min = -50;
> cfg.artfctdef.threshold.max = 50;
> cfg.artfctdef.prestim = -0.1;
> cfg.artfctdef.poststim = 0.6;
> cfg.artfctdef.window = 'trial';
> data_clean = ft_rejectartifact(cfg, clean_data);
>
> the call to "ft_preprocessing" took 111 seconds
> rejected 0 trials completely
> rejected 0 trials partially
> filled parts of 0 trials with NaNs
> filled parts of 0 trials with zeros
> filled parts of 0 trials with a specified value
>
>
> This is an example of what I get when using the functions together..
>
> cfg = [];
> cfg.continuous = 'no';
> cfg.artfctdef.channel = {'all', '-ECG1', '-ECG2', '-IO'};
> cfg.dataset =clean_data.trial(:,:);
> % cfg.artfctdef.threshold.min = -50;
> % cfg.artfctdef.threshold.max = 50;
> % cfg.artfctdef.threshold.range=100;
> cfg.artfctdef.prestim = -0.1;
> cfg.artfctdef.poststim = 0.6;
> cfg.trl=trl;
> [~, artData] = ft_artifact_threshold(cfg, clean_data);
>
> cfg = [];
> cfg.artfctdef.xxx.artifact = artData;
> cfg.artfctdef.reject = 'complete';
> fnData = ft_rejectartifact(cfg, clean_data);
>
> The output seems to not be able to read the trials information, as it does
> not mark the trials but only segments in which there are artifacts:
> searching for artifacts in 65 channels
> applying preprocessing options
> searching in trial 643 from 643
> detected 7172 artifacts
> the call to "ft_artifact_threshold" took 25 seconds
> >>
> cfg = [];
> cfg.artfctdef.xxx.artifact = artData;
> cfg.artfctdef.reject = 'complete';
> fnData = ft_rejectartifact(cfg, clean_data);
> detected 1 xxx artifacts
> Index in position 2 exceeds array bounds. Index must not exceed 1.
>
> Error in ft_rejectartifact (line 305)
> endsample = artifact{i}(:,2);
>
> Please help me!!
> Any support is appreciated
> Thank you very much
> *Eleonora Parrotta, Doctoral Researcher in Social Neuroscience *
> School of Psychology | University of Aberdeen | Aberdeen | AB24 3FX
> e.parrotta.20 at abdn.ac.uk
>
> _______________________________________________
> fieldtrip mailing list
> https://mailman.science.ru.nl/mailman/listinfo/fieldtrip
> https://urldefense.com/v3/__https://doi.org/10.1371/journal.pcbi.1002202__;!!HJOPV4FYYWzcc1jazlU!5HnRhfygr3VjQPF9taLrE4HIY5FtPbdMjJqQMfCezTF2khzpRnmVdA2Q-q9N34_LcdLAmMMuXm5FxQMa5rZfTBXCIjMTqw$
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20230525/dbd2b90a/attachment.htm>
More information about the fieldtrip
mailing list