How to avoid overwriting/saving the same plots in a Loop through multiple subjects across different conditions?

Hi everyone,

I have a dataset where I need to plot results of 20 subjects. My following code (converted from Brainstorm to Fieldtrip toolbox) first was saving time frequency (TF) plots for the 1st subject only in the results folder but not for other 19 subjects. I made some changes in the loop through subjects and it's saving the required plots for all subjects. But it is overwriting/ saving the same plot for all subjects.

Any suggestions where could be the mistake in the code? And how can I loop through so that it saves the plots for two conditions 1.Space 2. Time and 3. their difference Space-Time for each single subject?

New to matlab, your answer would be really appreciated.

Thanks in advance

Best,

Abrar

%%Time Frequency Representation
clear all
close all
clc

ft_folder = 'D:\fieldtrip-20200406\fieldtrip-20200406'
run([ft_folder 'ft_defaults.m'])
results_folder = 'D:\BS\wo\PICS';
mkdir(results_folder)
%% define folders
data_main = 'D:\BS\wo\data'; %data imported from Brainstorm software (BS)
subjects_id = 01:20
SubjectFolder = ls([data_main 'Subject*']) %SubjectFolder contains list of all subjects as folders Subject01, Subject02 and so on with .mat files
SubjectFolder = extractfield(dir([data_main 'Subject*', ]),'name');
folder_spatial_trials = 'Spatial_Correct';
folder_temporal_trials = 'Temporal_Correct';
% preparation struct
load('D:\BS\wo\code_extra\data_megin.mat')
cfg.layout = 'neuromag306all.lay';
fignum = 1
......................%some code
    %Following loop is saving the required plot but overwriting same plot for all subjects
for s = 1:lentgth(char(SubjectFolder))% loop through subjects for TF plots across space and time condition
    %The plot 1 which i need to save for each subject
    figure('Name',[char(SubjectFolder(s)) ' Space']), ft_multiplotTFR(cfg, TFRwave_space)
    set(gcf, 'units','normalized','outerposition',[0 0 1 1])
    saveas(gca,[results_folder '' char(SubjectFolder(s)) ' Space.jpeg']) % visualize and save Time-Freq plots for Spatial trials for each subject
end 

% The plot 2 which i need to save for each subject
for s = 1:lentgth(char(SubjectFolder))
    figure('Name',[char(SubjectFolder(s)) ' Time']), ft_multiplotTFR(cfg, TFRwave_time)
    set(gcf, 'units','normalized','outerposition',[0 0 1 1])
    saveas(gca,[results_folder '' char(SubjectFolder(s)) ' Time.jpeg']) % visualize and save Time-Freq plots for Temporal trials for each subject
end

TFRwave_diff = TFRwave_space; % visualize and save TF power difference for each subject
TFRwave_diff.powspctrm = (TFRwave_space.powspctrm - TFRwave_time.powspctrm)./(TFRwave_space.powspctrm + TFRwave_time.powspctrm);

% The plot 3 which i need to save for each subject
figure,
cfg = [ ] ;
% cfg.zlim = [- 1 1];
cfg.colorbar = 'yes';
cfg.layout = 'neuromag306all.lay';
ft_multiplotTFR(cfg, TFRwave_diff)
set(gcf, 'units','normalized','outerposition',[0 0 1 1])
saveas(gca,[results_folder '' char(SubjectFolder(s)) ' Space-Time.jpeg'])
end

[Sindar]

s = 1:lentgth(char(SubjectFolder))

• length, not lentgth. Is this your actual code?
• what is SubjectFolder? size and type (cell array, string array, something else?)
• converting to char before checking number of elements makes me nervous

[Get Cognitive]

Hi Sindar thanks for reply.

I have a list of 20 subjects with folders named as Subject01, Subject02,....,Subject20 with .mat files. SubjectFolder extracts this list. I was just trying different things with code so did this s = 1:lentgth(char(SubjectFolder)) which plotted Figs for all 20 subjects but it's saving same fig for all subjects

whos SubjectFolder
  Name               Size            Bytes  Class    Attributes

  SubjectFolder      1x20             2692  cell   
  

My actual code is this one

  %%Time Frequency Representation
clear all
close all
clc

ft_folder = 'D:\fieldtrip-20200406\fieldtrip-20200406\'
run([ft_folder 'ft_defaults.m'])
results_folder = 'D:\BS\wo\PICS\';
mkdir(results_folder)

%% define folders

data_main     = 'D:\BS\wo\data\';
subjects_id   = 01:20
SubjectFolder = ls([data_main 'Subject*'])
SubjectFolder = extractfield(dir([data_main 'Subject*']),'name');

folder_spatial_trials  =  'Spatial_Correct\';
folder_temporal_trials =  'Temporal_Correct\';

% preparation struct
load('D:\BS\wo\code_extra\data_megin.mat')
cfg.layout = 'neuromag306all.lay';
fignum = 1

%% computation

for s=1:size(SubjectFolder,1) % loop of subjects
    
    
    clear data_space data_time
    
    % prepare to build data strcut for space and time  condition
    trials_space = ls([data_main char(SubjectFolder(s)) '\' folder_spatial_trials 'data*trial*mat'])
    trials_time  = ls([data_main char(SubjectFolder(s)) '\' folder_temporal_trials 'data*trial*mat'])
    load([data_main char(SubjectFolder(s)) '\' folder_spatial_trials 'channel_vectorview306_acc1.mat'])
    labels       = extractfield(Channel,'Name');
    grad_idx     = setdiff(1:306, 3:3:306);
%     channel_idx  = grad_idx(ChannelFlag(grad_idx)~=-1);
    channel_idx  = grad_idx; % index of gradiometer
    label        = labels(channel_idx);

% create space struct
for tr = 1:size(trials_space,1)
    load([data_main char(SubjectFolder(s)) '\' folder_spatial_trials trials_space(tr,:)])
    data_space.trial{1,tr} = F(channel_idx,:);
    data_space.time{1,tr}  = Time;       
end

% create time struct
for tr = 1:size(trials_time,1)
    load([data_main char(SubjectFolder(s)) '\' folder_temporal_trials trials_time(tr,:)])
    data_time.trial{1,tr} = F(channel_idx,:);
    data_time.time{1,tr}  = Time;   
end

data_space.grad  = data_megin.grad;
data_time.grad   = data_megin.grad;
data_space.label = label';
data_time.label  = label';

%% 

%ch = 100
%tr = 56 
%figure, 
%plot(data_time.time{1,1}, data_time.trial{1,tr}(ch,:))
%hold on
%plot(data_space.time{1,1}, data_space.trial{1,tr}(ch,:))

%for ch = 1:204
    %for tr = 1:100
       % valmax_time(ch,tr) = max(abs(data_time.trial{1,tr}(ch,:)));
        %valmax_space(ch,tr) = max(abs(data_space.trial{1,tr}(ch,:)));
        

%% Compute Morlet Wavelet Time-Frequency (TF) Representations
cfg = [];
cfg.channel    = 'MEG';
cfg.method     = 'wavelet';
cfg.width      = 3;
cfg.output     = 'pow';
cfg.foi        = 1:2:45;
cfg.toi        = -6:0.05:12; %Epoch time (complete trial for correct trials only -6 to 12 seconds)
% cfg.toi        = -1:0.05:5;

TFRwave_space = ft_freqanalysis(cfg, data_space);
TFRwave_time = ft_freqanalysis(cfg, data_time);

%% Plot TF Results with baseline norm.
cfg = [];
cfg.baseline     = [-6 -5]; % TO CHANGE
cfg.baselinetype = 'absolute';
cfg.zlim         = [-5e-21 5e-21];
% cfg.xlim         = [-1 5];
cfg.showlabels   = 'no';
cfg.layout       = 'neuromag306all.lay';
cfg.colorbar     = 'yes';

figure, ft_multiplotTFR(cfg, TFRwave_space)
set(gcf, 'units','normalized','outerposition',[0 0 1 1])
saveas(gca,[results_folder '\'  char(SubjectFolder(s)) ' Space.jpeg'])

figure('Name',[char(SubjectFolder(s)) ' Time']), ft_multiplotTFR(cfg, TFRwave_time)
set(gcf, 'units','normalized','outerposition',[0 0 1 1])
saveas(gca,[results_folder '\'  char(SubjectFolder(s)) ' Time.jpeg'])


%
TFRwave_diff            = TFRwave_space;
TFRwave_diff.powspctrm  = (TFRwave_space.powspctrm - TFRwave_time.powspctrm)./(TFRwave_space.powspctrm + TFRwave_time.powspctrm); 
figure, 
cfg = [];
% cfg.zlim         = [- 1  1];
cfg.colorbar     = 'yes';
cfg.layout       = 'neuromag306all.lay';
ft_multiplotTFR(cfg, TFRwave_diff)
set(gcf, 'units','normalized','outerposition',[0 0 1 1])
saveas(gca,[results_folder '\'  char(SubjectFolder(s)) ' Space-Time.jpeg'])
end

[Stephen23]

Some notes:

• Use dir rather than ls. Its output is much easier to work with.
• Use fullfile rather than concatenating strings together. It is the correct tool for the job.
• load into an output variable, i.e. S = load(...) and access its fields.
• Align your code using the MATLAB Editor's default alignment rules (select all text, ctrl+i).
• Instead of superfluous type conversion char(SubjectFolder(s)) just use the correct indexing: SubjectFolder{s}
• To create different filenames in a loop use sprintf, just as the MATLAB documentation shows:

https://www.mathworks.com/help/matlab/import_export/process-a-sequence-of-files.html

[Get Cognitive]

Hi Stephen, I was just following a pipline in the Fieldtrip toolbox. Thank you for your recommendations, quite helpful I'll work on that. But any suggestions how can I cope with my current code to save the required plots for each subject?

[Sindar]

Assuming it isn't some weirdness with one of the above suggestions (I'm still eyeing that char suspiciously), what is the file it keeps overwriting? Is it what you'd expect for the first SubjectFolder? The last? Something completely different?

Also, can you copy the results of these:

SubjectFolder{1}
fprintf('%s\n',SubjectFolder{1:5})

Also, double check that you're actually looping. If SubjectFolder is really 1x20, then you only have one iteration here:

for s=1:size(SubjectFolder,1)

this is safer:

for s=1:numel(SubjectFolder)

[Get Cognitive]

Resolved makingt this correction, thanks!

for s=1:size(SubjectFolder,2)

instead of

for s=1:size(SubjectFolder,1) 
    or
for s = 1:lentgth(char(SubjectFolder))