Hugging Face's logo

Datasets:
jalauer
/
Gruendler2009

Modalities:
Text
Formats:
text
Size:
< 1K
Tags:
eeg
medical
clinical
classification
ocd
flanker
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
1
Gruendler2009 / scripts /Step2_GetERPs_REVISION_withERRORS.m
jalauer's picture
jalauer
Add files using upload-large-folder tool
b0cc889 verified
Raw History Blame Contribute Delete
13.8 kB
%% Step 2 - load pre-proc'd flankers data
clear all; clc
addpath('Z:\EXPERIMENTS\Flankers t2t\');
datalocation='Y:\EEG_Data\Flankers OCDII\Processed Data\'; % Data are here
savedir='Y:\EEG_Data\Flankers OCDII\ERPs\'; % Save processed data here
cd(datalocation);
% Load up
[xls_data,xls_hdr]=xlsread('Z:\EXPERIMENTS\Flankers t2t\info.xlsx');
% Data are 60 chans * 2000 time * N epochs | Ref'd to linked mastoids
FILZ=dir('*_FLANKERS.mat');
for si=length(FILZ):-1:1
subname=FILZ(si).name;
subno=str2num(subname(1:3)); clc
% Only run the latter on subjs pre-selected to be MUY BUENO (N=46)
if ~isempty(find(xls_data(:,1)==subno))
% Only run if not already done
if 1 % ~(exist([savedir,num2str(subno),'_ERPs.mat'])==2)
% Display
disp(['DOING: ',num2str(subno)]);
% Get Subj Info
tempdat=xls_data(find(xls_data(:,1)==subno),:);
tempdat_hdr=xls_hdr(find(xls_data(:,1)==subno)+1,:); % +1 b/c of header row
SUBJINFO(1)=subno;
SUBJINFO(2)=tempdat(1);
SUBJINFO(3)=tempdat(4);
SUBJINFO(4)=tempdat(6);
SUBJINFO(5)=tempdat(7);
SUBJINFO(6)=tempdat(9);
SUBJINFO_HDR={'EEG_ID','INFO_ID','Error%','Sex','Age','OCI_Score'};
%
if ~isnan(tempdat(5))
bad_ICAs_To_Remove=tempdat(5);
elseif isnan(tempdat(5))
bad_ICAs_To_Remove=str2num(cell2mat(tempdat_hdr(5)));
end
clear tempdat tempdat_hdr;
% Load EEG
load([datalocation,subname]);
% Remove the bad ICAs:
EEG = pop_subcomp( EEG, bad_ICAs_To_Remove, 0);
% Get the good info out of the epochs
VECTOR(1:size(EEG.epoch,2),1:13)=NaN;
for ai=1:size(EEG.epoch,2)
VECTOR(ai,1)=EEG.epoch(ai).TRIALNUM;
VECTOR(ai,2)=EEG.epoch(ai).STIM;
VECTOR(ai,3)=EEG.epoch(ai).CONGRU;
VECTOR(ai,4)=EEG.epoch(ai).RightLeft;
VECTOR(ai,5)=EEG.epoch(ai).RT;
VECTOR(ai,6)=EEG.epoch(ai).ACC;
% ^^^^^^^^^^^^^^^^^^ OOPS
if ~isempty(EEG.epoch(ai).PREVTRIAL)
% ^^^^^^^^^^^^^^^^^^
VECTOR(ai,7)=EEG.epoch(ai).PREVTRIAL.TRIALNUM;
VECTOR(ai,8)=EEG.epoch(ai).PREVTRIAL.STIM;
VECTOR(ai,9)=EEG.epoch(ai).PREVTRIAL.CONGRU;
VECTOR(ai,10)=EEG.epoch(ai).PREVTRIAL.RightLeft;
VECTOR(ai,11)=EEG.epoch(ai).PREVTRIAL.RT;
VECTOR(ai,12)=EEG.epoch(ai).PREVTRIAL.ACC;
VECTOR(ai,13)=EEG.epoch(ai).PREVTRIAL.BlockStart;
end
end
% Let's just do this for ERPs - keep EEG.data structure unfiltered
dims=size(EEG.data);
FILT.data=eegfilt(EEG.data,500,[],20);
FILT.data=reshape(FILT.data,dims(1),dims(2),dims(3));
% Set Params
tx=-2000:2:1998;
b1=find(tx==-200); b2=find(tx==0);
t1=find(tx==-500); t2=find(tx==1000);
N2topo1=find(tx==200); N2topo2=find(tx==350); N2toporangetot=200:2:350;
P3topo1=find(tx==400); P3topo2=find(tx==600); P3toporangetot=400:2:600;
tx2disp=-500:2:1000;
% Basecor your ERPs here so they are pretty.
BASE=squeeze( mean(FILT.data(:,b1:b2,:),2) );
for ai=1:dims(1)
FILT.data(ai,:,:)=squeeze(FILT.data(ai,:,:))-repmat( BASE(ai,:),dims(2),1 );
end
% $$$$$$$$$$$$$$$$$$$ Parse condis and save ERPs $$$$$$$$$$$$$$$$$$$ %
% If blockstart, then NaN the row
VECTOR(VECTOR(:,13)==1,[3,4,9,10])=NaN;
% % % If error, then NaN the row
% % VECTOR(VECTOR(:,6)==0,[3,4,9,10])=NaN;
% % % If post-error, then NaN the row
% % temp=VECTOR(:,6)==0;
% % VECTOR(logical([0;temp(1:end-1)]),[3,4,9,10])=NaN; clear temp;
% If RT<200 or NaN, then NaN the row
VECTOR(VECTOR(:,5)<200,[3,4,9,10])=NaN;
VECTOR(isnan(VECTOR(:,5)),[3,4,9,10])=NaN;
% If any >2, all are NaN
temp=sum((VECTOR(:,[3,4,9,10])>2)')';
VECTOR(temp>0,[3,4,9,10])=NaN;
% If any are NaN, all are NaN
temp=sum(isnan(VECTOR(:,[3,4,9,10]))')';
VECTOR(temp>0,[3,4,9,10])=NaN;
for ai=1:length(VECTOR)
if ~isnan(VECTOR(ai,3))
% ---------------------- congruent (1) or conflict (0)
if VECTOR(ai,9)==1 && VECTOR(ai,3)==1, GRATTON=1; % cC
elseif VECTOR(ai,9)==1 && VECTOR(ai,3)==0, GRATTON=2; % cI
elseif VECTOR(ai,9)==0 && VECTOR(ai,3)==1, GRATTON=3; % iC
elseif VECTOR(ai,9)==0 && VECTOR(ai,3)==0, GRATTON=4; % iI
end
% ---------------------- [Right=1, Left=2]
if VECTOR(ai,10)==1 && VECTOR(ai,4)==1, RESPS=1; % rR
elseif VECTOR(ai,10)==2 && VECTOR(ai,4)==2, RESPS=1; % lL
elseif VECTOR(ai,10)==1 && VECTOR(ai,4)==2, RESPS=10; % rL
elseif VECTOR(ai,10)==2 && VECTOR(ai,4)==1, RESPS=10; % lR
end
% ----------------------
VECTOR(ai,14)=GRATTON;
VECTOR(ai,15)=RESPS;
VECTOR(ai,16)=GRATTON.*RESPS;
clear GRATTON RESPS
end
end
UNIQUES=unique(VECTOR(:,16));
UNIQUES=UNIQUES(UNIQUES>0);
for ai=1:8
ERP(:,:,ai)=squeeze(mean(FILT.data(:,:,VECTOR(:,16)==UNIQUES(ai)),3));
end
% ---------------- single trial corr
for ai=1:8
for chani=1:60
rho=corr(squeeze(FILT.data(chani,:,VECTOR(:,16)==UNIQUES(ai)))', VECTOR(VECTOR(:,16)==UNIQUES(ai),5) ,'type','Spearman','rows','complete');
ERP_Corr(chani,:,ai)=rho;
end
end
% ---------------- limit based on lowest epoch count
for ai=1:8
TEMP_by_set{ai}=FILT.data(:,:,VECTOR(:,16)==UNIQUES(ai));
TEMP_by_set_SIZE(ai)=size(TEMP_by_set{ai},3);
end
TEMP_minsize=min(TEMP_by_set_SIZE);
% ----
for ai=1:8
TEMP_TRIALS=1:TEMP_by_set_SIZE(ai);
TEMP_TRIALS=shuffle(TEMP_TRIALS);
ERP_minsize(:,:,ai)=squeeze(mean( TEMP_by_set{ai}(:,:,TEMP_TRIALS(1:TEMP_minsize)) ,3));
clear TEMP_TRIALS
end
clear TEMP_by_set TEMP_by_set_SIZE TEMP_minsize TEMP_by_set_SIZE;
% ------------------ Now for response locked
for respi=1:dims(3)
RT=VECTOR(respi,5);
if isnan(RT)
response=zeros(dims(1),dims(2)) ; % repmat(NaN,dims(1),dims(2));
else
FILT_response=cat(2,FILT.data(:, floor(RT/2):end ,respi), zeros(dims(1),floor(RT/2)-1) );
response=cat(2,FILT.data(:, floor(RT/2):end ,respi), zeros(dims(1),floor(RT/2)-1) );
end
FILT.RESP(:,:,respi)=response;
EEG.RESP(:,:,respi)=response; clear FILT_response response;
end
% Stuff
L_C3=26;
R_C4=30;
for ai=1:length(VECTOR)
if ~isnan(VECTOR(ai,4))
% ---------------------- [Right=1, Left=2]
if VECTOR(ai,4)==1, active_site=L_C3; inactive_site=R_C4; % Right response, left cortex active
elseif VECTOR(ai,4)==2, active_site=R_C4; inactive_site=L_C3; % Left response, right cortex active
end
% ----------------------
FILT.RESP_ActInact(:,ai)=FILT.RESP(active_site,:,ai)-FILT.RESP(inactive_site,:,ai);
FILT.STIM_ActInact(:,ai)=FILT.data(active_site,:,ai)-FILT.data(inactive_site,:,ai);
else
FILT.RESP_ActInact(:,ai)=repmat(NaN,1,dims(2));
FILT.STIM_ActInact(:,ai)=repmat(NaN,1,dims(2));
end
end
% ERP-i-fy
for ai=1:8
ERP_RESP(:,ai)=squeeze(mean(FILT.RESP_ActInact(:,VECTOR(:,16)==UNIQUES(ai)),2));
ERP_RESP_All(:,:,ai)=squeeze(mean(FILT.RESP(:,:,VECTOR(:,16)==UNIQUES(ai)),3));
ERP_LRP_to_STIM(:,ai)=squeeze(mean(FILT.STIM_ActInact(:,VECTOR(:,16)==UNIQUES(ai)),2));
end
% ---------------- single trial corr
for ai=1:8
rho=corr(squeeze(FILT.RESP_ActInact(:,VECTOR(:,16)==UNIQUES(ai)))', VECTOR(VECTOR(:,16)==UNIQUES(ai),5) ,'type','Spearman','rows','complete');
ERP_RESP_Corr(:,ai)=rho;
end
% ---------------- limit based on lowest epoch count
for ai=1:8
TEMP_by_set{ai}=FILT.RESP_ActInact(:,VECTOR(:,16)==UNIQUES(ai));
TEMP_by_set_SIZE(ai)=size(TEMP_by_set{ai},2);
end
TEMP_minsize=min(TEMP_by_set_SIZE);
% ----
for ai=1:8
TEMP_TRIALS=1:TEMP_by_set_SIZE(ai);
TEMP_TRIALS=shuffle(TEMP_TRIALS);
ERP_LRP_minsize(:,ai)=squeeze(mean( TEMP_by_set{ai}(:,TEMP_TRIALS(1:TEMP_minsize)) ,2));
clear TEMP_TRIALS
end
clear TEMP_by_set TEMP_by_set_SIZE TEMP_minsize TEMP_by_set_SIZE;
% ^^^^^^^^^^^^^^^^ TF
% Setup Wavelet Params
num_freqs=50;
frex=logspace(.01,1.7,num_freqs);
s=logspace(log10(3),log10(10),num_freqs)./(2*pi*frex);
t=-2:1/EEG.srate:2;
% Definte Convolution Parameters
dims = size(EEG.data);
n_wavelet = length(t);
n_data = dims(2)*dims(3);
n_convolution = n_wavelet+n_data-1;
n_conv_pow2 = pow2(nextpow2(n_convolution));
half_of_wavelet_size = (n_wavelet-1)/2;
% same times, new baseline - also resp is -1000 to 500
tx=-2000:2:1998;
b1=find(tx==-300); b2=find(tx==-200);
t1=find(tx==-500); t2=find(tx==1000);
rt1=find(tx==-500); rt2=find(tx==1000); % OK, made this like cue to get lotsa post-response activity
% get FFT of data
TF_chani=19;
EEG_fft = fft(reshape(EEG.data(TF_chani,:,:),1,n_data),n_conv_pow2);
RESP_fft = fft(reshape(EEG.RESP(TF_chani,:,:),1,n_data),n_conv_pow2);
for fi=1:num_freqs
wavelet = fft( exp(2*1i*pi*frex(fi).*t) .* exp(-t.^2./(2*(s(fi)^2))) , n_conv_pow2 ); % sqrt(1/(s(fi)*sqrt(pi))) *
% convolution
EEG_conv = ifft(wavelet.*EEG_fft);
EEG_conv = EEG_conv(1:n_convolution);
EEG_conv = EEG_conv(half_of_wavelet_size+1:end-half_of_wavelet_size);
EEG_conv = reshape(EEG_conv,dims(2),dims(3));
% convolution
RESP_conv = ifft(wavelet.*RESP_fft);
RESP_conv = RESP_conv(1:n_convolution);
RESP_conv = RESP_conv(half_of_wavelet_size+1:end-half_of_wavelet_size);
RESP_conv = reshape(RESP_conv,dims(2),dims(3));
% Get baseline
BASE = mean(mean(abs(EEG_conv(b1:b2,:)).^2));
for ai=1:8
% Get power by condi
temp_POWER1 = mean(abs(EEG_conv(t1:t2,VECTOR(:,16)==UNIQUES(ai))).^2,2);
temp_POWER2 = mean(abs(RESP_conv(rt1:rt2,VECTOR(:,16)==UNIQUES(ai))).^2,2);
% dB correct power by base (different time windows)
POWER{1}(fi,:,ai) = 10*( log10(temp_POWER1) - log10(repmat(BASE,size(tx2disp,2),1)) ) ;
POWER{2}(fi,:,ai) = 10*( log10(temp_POWER2) - log10(repmat(BASE,size(tx2disp,2),1)) ) ;
% Get ITPC by condi (different time windows)
ITPC{1}(fi,:,ai) = abs(mean(exp(1i*( angle(EEG_conv(t1:t2,VECTOR(:,16)==UNIQUES(ai))) )),2));
ITPC{2}(fi,:,ai) = abs(mean(exp(1i*( angle(RESP_conv(t1:t2,VECTOR(:,16)==UNIQUES(ai))) )),2));
clear temp_POWER1 temp_POWER2;
end
clear wavelet EEG_conv RESP_conv BASE;
end
% ^^^^^^^^^^^^^^^^
save([savedir,num2str(subno),'_ERPs_2018_Revision_withErrors.mat'],'SUBJINFO','SUBJINFO_HDR','ERP','ERP_RESP','ERP_RESP_All','VECTOR','POWER','ITPC',...
'ERP_LRP_to_STIM','ERP_minsize','ERP_Corr','ERP_RESP_Corr','ERP_LRP_minsize');
end
end
clearvars -except FILZ xls_data xls_hdr si datalocation savedir;
end
%%