There were a few steps to making the nii files:

• Finding the right values for the convert shell file.
• Fixing/debugging errors and changing err/out files.
• Finding what to change in the heuristic file.
• Finding appropriate values for the heuristic file.

• Using the default converter (no -c means dcm2nii), the file would start like this: ./dicomconvert2_GE.py -d /scratch/... -o /scratch/... -f /scratch/... -q ... -s ...

• This step involved looking into the dicomconvert2_GE.py file to double check what the parameters were.

• -d is the dicom folder.

• -o is the output project folder.

• -f is the heuristic file that establishes what types of scans for which nii files are created.

• -q is the LSF queue.

• -s is the subject list (one person).

• -c was the converter, but I'm using the default (no known reason to use mri_convert over dcm2nii yet).

• Thus, the file ended like this: ./dicomconvert2_GE.py -d /scratch/PSB6351_2017/dicoms -o /scratch/PSB6351_2017/week4/hays/project_folder/ -f /scratch/PSB6351_2017/week4/hays/heuristic_shell.py -q PQ_fasoto -s subj001

• Running the convert_dicom_shell.sh as it was would cause both runtime errors and debugging issues.
• The variable, trd_bold, was not defined and needed to simply be "bold" in the heuristic file (because "bold" was defined and the placement corresponded with the t1 and dwi).
• Furthermore, the err and out file locations needed to be updated in the dicomconvert2_GE.py to avoid overriding other students' err and out files.

• I wrote debugging file writers (because print() doesn't always work conveniently on LSF jobs) to figure out what values seqinfo contained before I knew what the heuristic file parameters were.
• They suggested that sl was the total slice count, nt was number of temporal positions, and s[12] was the series description.

• To find the slices for the structural scan, you can check Z01 (a T1 dicom) for the slice count (186) using the dicom_hdr command and the bash command grep with Perl regular expressions. dicom_hdr /scratch/PSB6351_2017/dicoms/subj001/A/Z01 | grep -Po .*Acquisition//.*

• There was no particular reason to prevent the bold dicoms' nt's from being a single value, so it was left as nt != "**" because it would not affect any of them -- the sub-name restriction of "TRD" was enough to get the full list of bold dicom files.

• To make sure there were not negative nt values (or some other obvious bad values). The full list of the number of temporal positions was obtained with: dicom_hdr /scratch/PSB6351_2017/dicoms/subj001/A/* | grep -Po .*Temporal.* | sort | uniq

• To find a list of all applicable scan descriptions, the following command can be run to pipe the header to a regular expression search: dicom_hdr /scratch/PSB6351_2017/dicoms/subj001/A/* | grep -Po .*Descr.* | sort | uniq

• This revealed that the diffusion weighted image needed to look for "DTI" (after waiting a while!), which represents diffusion tensor imaging.

• In total, there were 7 bold nii files, 1 diffusion weighted image nii file, and 1 anatomical (T1) nii file, which were readable in afni's viewer.