I went through the FSL-MRS practical with the example data and it worked perfectly. Now I am trying to use my own data (Siemens, TWIX), but I am running into an issue with the coil combination. It says:
ValueError: operands could not be broadcast together with shapes (2080,2,49,32,1,1,1) (2,32,1,1,1)
In the attatchment below is shown the info of these files and at the very last line the command I tried to use.
I don’t understand why it shows in the error message that the shape of the wref_temp is only (1,1,1,32,2). Where has the value 2080 gone? Is this the problem?
And also when comparing these to the example data it looks like the ‘DIM_USER_0’ and the ‘DIM_DYN’ are the wrong way around. I am guessing that the DIM_DYN is supposed to be the 2 and the ‘DIM_USER_0’ to be the 49. How can I fix this?
Also I am guessing that the DIM_USER_0 is the amount of FIDs in this file. Do I need to get rid of that dimension also and if yes, how do I do that?
Looking into it a little more I understand this better now. I still dont know what the DIM_USER_0 is, it looks just has an averaged spectra and then a spectra that is 0.
Therefore
1.Not a problem, but still don’t know why it isn’t shown.
2. it is not flipped. it is as it is supposed to be
3. I am guessing, that you can get rid of the extra dimension with just averaging the dimension with:
fsl_mrs_proc average --file data/FILE --dim DIM_USER_0 --output OUTPUTFOLDER --filename NEWFILE
Just be sure to do this to both water reference and metabolite data and after that start the preprocessing accordingly.
Sorry for the delay in replying. The first error arises from the water references having two averages. You can fix this by running fsl_mrs_proc average --file wref.nii.gz --dim DIM_USER_0 --output .
I should adjust the error output to be more helpful for this. The DIM_USER_0 tag is just a generic tag when spec2nii doesn’t have any information on what the dimension is used for. DIM_DYN is the number of repeats.
The empty spectrum is a bit odd. What sequence is this from?
