(subjectively) IMPORTANT QUESTION! Please help.

Hi,

I am creating a .m file and a GUI for quality assurance purposes to inspect segmented ROIs previously worked on by a student. The data is of mice and is organized by weeks. For example, Week 0 would have 9 mice associated with that data set. The workflow that I made (although I am sure it will be criticized) is to load the experimental mice data into a struct into MATLAB. This is done per week. So what I'd ideally like to have is 4 structs (W0, W4, W8, W20). I would like to then save this struct so that I can use it at a later date. Because the entire process takes quite some time (~8 minutes per mouse, with 9 mice per week), I think this would be helpful. The problem that I am having is that, I receive an "OUT OF MEMORY" error from MATLAB after going through 3 mice. As you can tell, this destroys my aforementioned workflow since I'd like to save an entire struct containing specifically chosen data for all mice imaged at a certain week.

The "quality assurance" is accomplished through my creation of a GUI that enables the plotting of the data from the 3 acquired positions per mouse, per frame, per week for easy inspection of the segmented ROIs.

I am currently running a main script that: 1. Loads into MATLAB multiple single files that contains experimental data from individual mice (each file corresponding to a single mouse). 2. Then, I create a struct in MATLAB that first contains fields with the names of the corresponding mice and other respective data. 3. Then, I create more structs inside this struct the data that I need to pass into the struct. The over-arching struct used is called 'mouseinfo'. 4. Then, I pass 'mouseinfo' into my GUI and everything else works wonderfully. The issue lies in Step 3. with passing all the experimental mouse data into the struct.

See this major snippet of my main code:

% 
  mouseinfo(1).cage = 'B140308';
% mouseinfo(2).cage = 'B140536';
% mouseinfo(3).cage = 'B140874';
% mouseinfo(4).cage = 'LL140536';
% mouseinfo(5).cage = 'R140308';
% mouseinfo(6).cage = 'R140536';
% mouseinfo(7).cage = 'R140874';
% mouseinfo(8).cage = 'RR140536';
% mouseinfo(9).cage = 'RR140874';

    for n = 1%9
    mouseinfo(n).week = 0;
    mouseinfo(n).cage_nr = 1;
    ii = 1;
    col = 'rgb';
    for pos = 0:2    
        Output_fname = strcat('Outputs_', mouseinfo(n).cage,'_W', num2str(mouseinfo(n).week), '_left_spectro', num2str(pos));
        OP = load([Output_fname, '.mat']);
        OP = OP.(Output_fname);

        mouseinfo(n).pos(pos+1).ROI_mask = OP.ROI_mask;
        mouseinfo(n).pos(pos+1).ROI_xi = OP.xi;
        mouseinfo(n).pos(pos+1).ROI_yi = OP.yi;
        mouseinfo(n).pos(pos+1).PA_avg = OP.PA_avg;
        mouseinfo(n).pos(pos+1).US_avg = OP.US_avg;
        mouseinfo(n).pos(pos+1).PA_fit = OP.PA_fit;
        mouseinfo(n).pos(pos+1).US_fit = OP.US_fit;
        mouseinfo(n).pos(pos+1).PA_slope = OP.PA_slope;
        mouseinfo(n).pos(pos+1).US_slope = OP.US_slope;

        jj = 1;
        for frame = 1:59
            clear PA_fname
            frame
            ENW = xlsread(strcat(mouseinfo(n).cage,'_W', num2str(mouseinfo(n).week), '_left_spectro', num2str(pos), '_PAMODE.csv'));
            mouseinfo(n).pos(pos+1).frame(frame).ene = mean(ENW(frame+7, 8:11));

            % Load PA data
            PA_fname = strcat('PA_', mouseinfo(n).cage,'_W', num2str(mouseinfo(n).week), '_left_spectro', num2str(pos), '_fr_', num2str(frame));
            PA = load([PA_fname, '.mat']);
            PA = PA.(PA_fname);

            mouseinfo(n).pos(pos+1).frame(frame).PA = real(PA)./mouseinfo(n).pos(pos+1).frame(frame).ene;
            [pa_row, pa_col] = size(mouseinfo(n).pos(pos+1).frame(frame).PA);
            [mouseinfo(n).pos(pos+1).frame(frame).PA_SA, mouseinfo(n).pos(pos+1).frame(frame).PA_LCI] = compute_SA_LCI(mouseinfo(n).pos(pos+1).frame(frame).PA);

            % Load US data
            US_fname = char(strcat(mouseinfo(n).cage,'_W', num2str(mouseinfo(n).week), '_left_spectro', num2str(pos), '.iq'));
            [mouseinfo(n).pos(pos+1).frame(frame).US, mouseinfo(n).pos(pos+1).frame(frame).LatD, mouseinfo(n).pos(pos+1).frame(frame).AxD] = US_VevoLAZR_ReconstructRF(US_fname, 1, box_start, box_end, box_width, pa_row, pa_col);
            [mouseinfo(n).pos(pos+1).frame(frame).US_SA, mouseinfo(n).pos(pos+1).frame(frame).US_LCI] = compute_SA_LCI(mouseinfo(n).pos(pos+1).frame(frame).US);
              mouseinfo(n).pos(pos+1).frame(frame).PA2 = (mouseinfo(n).pos(pos+1).frame(frame).PA) .* (mouseinfo(n).pos(pos+1).ROI_mask);
             [mouseinfo(n).pos(pos+1).frame(frame).PA2_SA, mouseinfo(n).pos(pos+1).frame(frame).PA2_LCI] = compute_SA_LCI(mouseinfo(n).pos(pos+1).frame(frame).PA2);
             mouseinfo(n).pos(pos+1).frame(frame).PA2_SA_avg= nanmean(mouseinfo(n).pos(pos+1).frame(frame).PA2_SA(:));
%             
             mouseinfo(n).pos(pos+1).frame(frame).PA2 = remove_nan(mouseinfo(n).pos(pos+1).frame(frame).PA2);
             [pa_row1, pa_col1] = size(mouseinfo(n).pos(pos+1).frame(frame).PA2);

            % FFT
            [pa_freq, mouseinfo(n).pos(pos+1).frame(frame).PA_PS] = sig_fft(mouseinfo(n).pos(pos+1).frame(frame).PA2, pa_fs, 2);
            mouseinfo(n).pos(pos+1).frame(frame).PA_PS_avg = nanmean(mouseinfo(n).pos(pos+1).frame(frame).PA_PS, 2);
            mouseinfo(n).pos(pos+1).frame(frame).PA_PS_fr = mouseinfo(n).pos(pos+1).frame(frame).PA_PS_avg;     
            [freq_ind, mouseinfo(n).pos(pos+1).frame(frame).ISP] = int_spec_pow(pa_freq, mouseinfo(n).pos(pos+1).frame(frame).PA_PS_avg);

        end
        clear Output_fname
        clear OP
    end
end

Once again, my issue is that I receive an "Out of Memory" error from MATLAB after 3 mice after loading in specific data into the struct 'mouseinfo'. Any help is appreciated! I'd love to have this solved ASAP.

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/29922-why-your-question-is-not-urgent-or-an-emergency

[Walter Roberson]

Now, although it is true that I have been known to post unimportant questions, because I am a curious completist, it seems to me that most people who post here consider their questions to be "important", at least to them.

Answer 1

[Stephen23]

"The problem that I am having is that, I receive an "OUT OF MEMORY" error from MATLAB after going through 3 mice."

Basically you can either:

• buy more memory, or
• change your process.

"As you can tell, this destroys my aforementioned workflow since I'd like to save an entire struct containing specifically chosen data for all mice imaged at a certain week."

Yeah, it would be nice if computers had infinite memory and ran infinitely fast, then I could also implement some pretty neat algorithms.

If you want to get this going, then obviously you will have to change your process. Here are some concrete suggestions of ways to move forward:

• Use matfile instead of reading the data into memory. Perhaps for both reading and writing.
• Read one file at a time and collect one field/whatever. Save. Repeat for each field that you need.
• Your data names indicate that you have image/video data, which is inevitably a huge memory eater. Consider using the video processing tools, which are designed to help manage that.
• Check that you are not duplicating any data in memory. Avoid indexing/slices/splitting up arrays.