problem for brace in array name

Hello, I have a code as followings.

% Check your directory of SPM 
addpath('C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\spm12'); 
spm; 
% Close the pop-up window. 


% Check the directory where you saved the dataset 
dir_name = 'C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\data';
group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 
G = 3; 

% Load the data set. 
X = []; 
for g = 1:G
    file_name = dir([dir_name group_name{g}]);
    count = 1; 
    for i = 1:length(file_name)
        tname = [dir_name group_name{g} '\' file_name(i).name];
        if ~isempty(strfind(tname,'img'))
            test = spm_vol(tname); 
            X{g}(:,:,:,count) = spm_read_vols(test);
            count = count + 1; 
        end
    end
end
% control data is in X{1}. 
size(X{1})
% 91*109*91*11 : 91*109*91 images, 44 subjects

% MCI data is in X{2} 
size(X{2}) 
% 91*109*91*24 : 91*109*91 images, 24 subjects

% AD data is in X{3} 
size(X{3}) 
% 91*109*91*22 : 91*109*91 images, 22 subjects

save 'X.mat' 'X'; 

I want to save image files into array type.

So, I defined array names using brace indexing.

But, an error occurs after running this code.

Brace indexing is not supported for variables of this type.

Error: test_load_dataset (line 34)

size(X{1})

What should I do? T.T

I am very very beginner for MATLAB. T.T

Please give me hands.

Thanks.

[KSSV]

It seems X is a cell.....can you tell what

whos X

outputs?

[Hyunjong Lee]

It is a group from each dataset.

An output from image files.

[KSSV]

I asked you something else......what

whos X

outputs?

[Hyunjong Lee]

Oh sorry

it says as followings.

  Name      Size            Bytes  Class     Attributes

  X         0x0                 0  double    

[Bjorn Gustavsson]

Before the look you define X:

X = [];

That creates X as a normal double array with size 0 x 0, then you try to treat X as a cell-array, which fails. Change your initialization to:

X = {};

(Whether necessary or not it might help against workspace cluttering if you want to run multiple similar scripts where X is used as a target variable.)

HTH

[Stephen23]

"Change your initialization to: X = {};"

That will not resolve the root problem, which is that the code cannot handle the case when the IF is not true and so nothing is allocated to X (the fact the error occurs after the loops tells us this is actually the case). That change will simply mean errors due to accessing X{3}, etc.

A robust solution is to preallocate the cell array correctly:

X = cell(1,G);

and then detect the empty cells after the loop.

[Guillaume]

While we're at it:

group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 
G = 3; 
for g = 1:G
  ...

So, if you add or remove groups, you need to edit group_name and G and make sure that they match. It's much safer to ask matlab how many groups there are and use that value:

group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 
for g = 1:numel(group_name)

It's also less lines of code.

[KSSV]

Try this once.

% Check your directory of SPM 
addpath('C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\spm12'); 
spm; 
% Close the pop-up window. 
% Check the directory where you saved the dataset 
dir_name = 'C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\data';
group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 
G = 3; 
% Load the data set. 
X = cell(G,1); 
for g = 1:G
    file_name = dir([dir_name group_name{g}]);
    count = 1; 
    for i = 1:length(file_name)
        tname = [dir_name group_name{g} '\' file_name(i).name];
        if contains(tname,'img')
            test = spm_vol(tname); 
            X{g} = spm_read_vols(test);            
        end
    end
end
% control data is in X{1}. 
size(X{1})
% 91*109*91*11 : 91*109*91 images, 44 subjects
% MCI data is in X{2} 
size(X{2}) 
% 91*109*91*24 : 91*109*91 images, 24 subjects
% AD data is in X{3} 
size(X{3}) 
% 91*109*91*22 : 91*109*91 images, 22 subjects
save 'X.mat' 'X'; 

[Hyunjong Lee]

Thank you for your help.

I ran the code as followings.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Load the dataset. 
% You need to download SPM toolbox from 
% http://www.fil.ion.ucl.ac.uk/spm/software/

% Check your directory of SPM 
addpath('C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\spm12'); 
spm; 
% Close the pop-up window. 


% Check the directory where you saved the dataset 
dir_name = 'C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\data';
group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 

% Load the data set. 
X = cell(1,numel(group_name)); 
for g = 1:numel(group_name)
    file_name = dir([dir_name group_name{g}]);
    count = 1; 
    for i = 1:length(file_name)
        tname = [dir_name group_name{g} '\' file_name(i).name];
        if ~isempty(strfind(tname,'img'))
            test = spm_vol(tname); 
            X{g}(:,:,:,count) = spm_read_vols(test);
            count = count + 1; 
        end
    end
end


% control data is in X{1}. 
size(X{1})
% 91*109*91*11 : 91*109*91 images, 44 subjects

% MCI data is in X{2} 
size(X{2}) 
% 91*109*91*24 : 91*109*91 images, 24 subjects

% AD data is in X{3} 
size(X{3}) 
% 91*109*91*22 : 91*109*91 images, 22 subjects

save('X.mat','X'); 

But I could not get the right answer.

It says,

ans =

     0     0


ans =

     0     0


ans =

     0     0

There is no value in all cells T.T

[Bjorn Gustavsson]

Then that might be because you never have the condition in your if-statement to be true? Check that - that is run your code step by step for g = 1, and see that you actually get what you expect out of that.

Answer 1

[Guillaume]

Clearly, if each cell is empty, that's because no file is found. Perhaps you build your path wrong, or perhaps there is no file there. In any case, it's trivial to include a check and a message that warns you if no file is found so at least you can check that you've got the correct path.

If I understood your code correctly, you're looking for any file with img in their name within the group directory. You can ask for these directly with dir avoiding the strfind:

dir_name = 'C:\Users\hyunj\Dropbox\nm_brain_connectivity_tutorial\data';
group_name = {'9110991_NC_cerebnor','9110991_QD_cerebnor','9110991_AD_cerebnor'}; 

group_vols = cell(1, numel(group_name));  %much better variable name than X!
for g = 1 : numel(group_name)
    groupfolder = [dir_name, group_name{g}];
    filelist = dir(fullfile(groupfolder, '*img*'));  %find all files containing img in groupfolder
    if isempty(filelist)
        warning('No file found in directory: %s', groupfolder);
    else
        vols = arrayfun(@(file) spm_read_vols(spm_vol(fullfile(groupfolder, file.name))), filelist, 'UniformOutput', false);
        group_vols = cat(4, vols{:});
    end
end

Note that instead of growing X{g} in a loop using a count variable, I use arrayfun to generate a cell array that is then converted into a 4D matrix.

[Hyunjong Lee]

Thank you for all.

Now I've understood what the problem was.

Thank you so much everybody ^^.