Dear Alpaslan,

Assuming the name of the image file is "med.dat", you first load the image in the matlab working environment. That is: load med.dat. Then, observe the variables contained in the file, and the commands for this are "who" and "whos" (you simply type  "whos", for example, in the command window). Assume the interest variable that contains the slices is "slice" (which should be 128*128*134--the size of your image-volume), then you can extract the i-th frame by slice(:,:,i); the first slice is slice(:,:,1). Putting this in a for-loop you can extract automatically all frames this way:

%% Prepare a variable to store your results.

stor = zeros(128,128,374)

for i=1:m

............

image-i=slice(:,:,i);

............% some operations on the i-th image

stor(:,:,i)=processed_image-i

end

save stor

%%%%%%%%%%%%%%

With the above 'skeleton' of the code you can also save the processed frames back to the single file.

As mentioned above, if xxx.dat file contains only data, not the header information, you can read it using "fread" matlab function. Check format type of data, e.g. short, long, or float.

To generate a dicom file, dicomwrite can be used. 

To generate a nifti file in Matlab, you can "spm_write_vol" in SPM library or "WriteBrik" from AFNI matlab libary and convert BRIK file to nifti file later using 3dcopy xxx+orig xxx.hdr. 

Hope it helps!

its better if you convert this file to a known image format, then perform your deconvolution process.