Albrecht2019 / Pre-processing_Pseudo.txt
jalauer's picture
Add files using upload-large-folder tool
69ebf88 verified
%% Pre-processing pipeline pseudocode
%% Everything run in MATLAB
%% Necessary extensions - eeglab, preppipeline, AMICA, corrmap, CSD
% Preprocessing
read in data
re-reference to average
remove baseline
insert channel locations 'MMN_ElectrodeCoords_MA2.ced)'
High Pass Filter - 'pop_eegfiltnew( EEG, 1, [], [], [0], 0, 0, 0);'
Run preppipeline
intpChans = setxor(1:64, [5 27 38 60]); % which channels to use for interpolation; exclude EOG channels
params = struct('name', [vhdr{1} 'snum' num2str(sx) '.' num2str(nf)], ...
'lineFrequencies', [60:60:300], ... % Line frequencies to remove
'referenceChannels', intpChans, ... % Channels to be used for re-referencing
'evaluationChannels', intpChans, ... % Channels to be used for interpolation
'rereferencedChannels', [1:64], ... % Channels to be linenoise removed and referenced
'detrendChannels', [1:64], ... % Channels to detrend
'lineNoiseChannels', [1:64], ... % Channels to remove line noise from
'ignoreBoundaryEvents', true, ... % ??? not sure why it doesn't work without
'detrendType', 'none', ... % Use high pass filter to detrend channels
'detrendCutoff', 1, ... % High pass filter frequency cut off
'referenceType', 'robust', ... % Robust reference
'meanEstimateType', 'median', ... % Use median for robust reference
'interpolationOrder', 'post-reference', ... % Interpolate channels after robust referencing
'keepFiltered', true); % Retain filtering?
EEG = prepPipeline(EEG, params);
Low Pass Filter -
pop_eegfiltnew( EEG, [], 50, [], [0], 0, 0, 0);
Epoch to -1.5 - 3 seconds around stimulus using these triggers
[111:114 121:124 211:214 221:224]
Remove large artifacts using pop_autorej
pop_autorej(EEG, 'nogui','on','threshold',500, 'startprob',5,'maxrej',5);
Downsample to 500 Hz
pop_resample( EEG, 500 );
Run AMICA using the following
[EEG.icaweights, EEG.icasphere, EEG.icamods] = runamica15(EEG.data(:,:), ...
'max_iter', 10000, 'writestep', 500, 'pcakeep', 42, 'do_reject', 1, ...
'max_threads', 12, 'rejsig', 4);
Create eeglab 'STUDY' structure
Precompute topography to extract EOG using STUDY
std_precomp(STUDY, ALLEEG, 'components','allcomps','on','recompute','on','scalp','on');
Find prototypical VEOG artifact and run 'corrmap' to identify all VEOG components
corrmap(STUDY, ALLEEG, indx, 1, 'ics',1,'pl', 'both','clname','VEOG','badcomps','no', 'resetclusters','on');
Reject epochs by trend and component spectra, exclude for VEOG artifact
pop_rejtrend( EEG, 0, include, size(EEG.times,2), 5, 0.3, 0, 0, 0);
pop_rejspec( EEG, 0, 'elecrange', include ,'threshold',[-50 45;-100 30],'freqlimits',[0 3;20 50],'eegplotcom','','eegplotplotallrej',0,'eegplotreject',0);
EEG.reject.sum = (EEG.reject.icarejfreq + EEG.reject.icarejconst) > 0;
pop_select( EEG, 'notrial', find(EEG.reject.sum));
Run second round ICA on cleaner data
[EEG.icaweights, EEG.icasphere, EEG.icamods] = runamica15(EEG.data(:,:), ...
'max_iter', 10000, 'writestep', 500, 'pcakeep', 42, 'do_reject', 1, ...
'max_threads', 12, 'rejsig', 4);
Identify VEM, HEOG and VEOG artifacts and run corrmap again to identify
Subtract artifacts from EEG
pop_subcomp( EEG, [VEM HEOG VEOG], 0); % Check this
Final clean up - re-baseline correct to 100 ms epoch baseline and remove large noisy epochs
pop_rmbase( EEG, [-100 0]);
pop_eegthresh( EEG, 1, [1:64], -90, 90, -1.5, 3, 0, 1);
Run CSD transformation
Remove eye channels
ConvertLocations
ExtractMontage for CSD
Check Montage (MapMontage(M))
Set CSD: [G,H] = GetGH(M, 4);
Run CSD: CSD(EEG.data(:,:), G, H, 1.0e-5, 10);
reshape(EEG.CSDdata, size(EEG.data));