[FieldTrip] something wrong with my permutation test for ERP

[蔡林]

Dear fieldtripers,

I have already understood the tuitorials about how to do a permutation test for ERP analysis.
I made some codes to do that. But I got some strange results and plots. And I couldnot find problems about my codes.

%%%%%%%%%%%Codes:

cfg = [];
cfg.keepindividual = 'yes';
cfg.channel     = 'all';
avg_12 = ft_timelockgrandaverage (cfg, data_12(:).ERP);
avg_22 = ft_timelockgrandaverage (cfg, data_22(:).ERP);
clear data*;

outfil = strcat('/EEG/Discourse_Exp2/n16_grandavg_keeptrial_all_12vs22');
save(outfil, 'avg_12', 'avg_22');

load /EEG/Discourse_Exp2/n16_grandavg_keeptrial_all_12vs22;
load (sprintf('E:/EEG/Discourse_Exp2/neighbours_Lin.mat'));
% cfgneigh.neighbourdist = 42; %or 45, define the cluster neighbours % select all channels within 40 mm distance of the current channel as neighbours
% cfgneigh.elec = elec; % read channel locations and labels from this file
cfg.neighbours       = neighbours_build; 

% load J:/new_4_names/data_valence/fieldtrip/elec_60;
% cfgneigh.neighbourdist = 42; % select all channels within 36 mm distance of the current channel as neighbours
% cfgneigh.elec = elec; % read channel locations and labels from this file
% cfg.neighbours       = ft_prepare_neighbours(cfgneigh); 

cfg.channel          = {'all'};
cfg.method = 'montecarlo';
% cfg.design           = [1:24 1:24; ones(1,24), ones(1,24) * 2];
cfg.design           = [1:16 1:16; ones(1,16), ones(1,16) * 2];
cfg.uvar             = 1; % subject number (unit variable) on line 1 of the design matrix
cfg.ivar             = 2; % condition number (independent variable) on line 2 of the design matrix

cfg.latency = [0.2 1];
cfg.avgovertime = 'no';%(default = 'no')

cfg.numrandomization = 1000;
cfg.correctm         = 'cluster';
cfg.alpha            = 0.05;
cfg.tail             = 0; % one-or two-sided testing
cfg.clusterstatistic = 'maxsum'; % maximum sum of t-values within one cluster is the test statistic
cfg.clusterthreshold = 'parametric'; % paired-sample t-test for the uncorrected t-values
cfg.clusteralpha     = 0.05;
cfg.clustertail      = 0; % two-sided testing;
cfg.statistic = 'depsamplesT';

statis_all_12vs22 = ft_timelockstatistics(cfg, avg_12, avg_22);
outfil = strcat('/EEG/Discourse_Exp2/n16_statis_all_12vs22');
save(outfil, 'statis_all_12vs22')

%%%%%clustor plot
load /EEG/Discourse_Exp2/n16_grandavgERP_resp;
load /EEG/Discourse_Exp2/n16_statis_all_12vs22;
GA_RvsC = grandavg_12;
GA_RvsC.avg = grandavg_22.avg - grandavg_12.avg;

figure;  
timestep = 0.05;      %(in seconds)
sampling_rate = 500;
sample_count = length(statis_all_12vs22.time);
j = [0:timestep:1];   % Temporal endpoints (in seconds) of the ERP average computed in each subplot
m = [1:timestep*500:sample_count];  % temporal endpoints in MEEG samples

pos_cluster_pvals = [statis_all_12vs22.posclusters(:).prob];
pos_signif_clust = find(pos_cluster_pvals < statis_all_12vs22.cfg.alpha);
pos = ismember(statis_all_12vs22.posclusterslabelmat, pos_signif_clust);

neg_cluster_pvals = [statis_all_12vs22.negclusters(:).prob];
neg_signif_clust = find(neg_cluster_pvals < statis_all_12vs22.cfg.alpha);
neg = ismember(statis_all_12vs22.negclusterslabelmat, neg_signif_clust);

pos = statis_all_12vs22.posclusterslabelmat == 1;	% or == 2, or 3, etc.
neg = statis_all_12vs22.negclusterslabelmat == 1;

for k = 1:16;
     subplot(4,4,k);   
     cfg = [];   
     cfg.layout = 'Lin_use.lay';
     cfg.xlim=[j(k) j(k+1)];   
     %cfg.zlim = [-1.0e-13 1.0e-13];   
     pos_int = all(pos(:, m(k):m(k+1)), 2);
     neg_int = all(neg(:, m(k):m(k+1)), 2);
     cfg.highlight = 'on';
     cfg.highlightchannel = find(pos_int | neg_int);       
     cfg.comment = 'xlim';   
     cfg.commentpos = 'title';   
     ft_topoplotER(cfg, GA_RvsC);
end

Please help me.

Thanks a lot!
-- 
Lin Cai
Department of Psychology, Peking University, Beijing 100871, P.R.China
-------------- next part --------------
A non-text attachment was scrubbed...
Name: strange2.jpg
Type: image/jpeg
Size: 150724 bytes
Desc: not available
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140116/e0262f21/attachment-0002.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: strange.jpg
Type: image/jpeg
Size: 49512 bytes
Desc: not available
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20140116/e0262f21/attachment-0003.jpg>