neurospin / pypreclin Goto Github PK

Hi people,

Congrats on getting the pipeline out and thanks for making it all open. I am interested in comparing the results against our/my own preprocessing pipeline (https://github.com/VisionandCognition/NHP-BIDS) to see if it would make sense to change our workflow. I'll be trying some things in the upcoming days/weeks depending on available time and will report back here with anything that's unclear, won't work or required extra tweaking.

To start things off:

• there's a typo in the very first 'functional' line of your example preproc.py (and the notebook too). It tries to import prepoc instead of preproc, which of course doesn't work. It took me a while to not misread it as it should be and realize the 'r' was missing.
• os is not imported, but it is used
• the way the example is set up, warp may be set to False, but the arguments to do unwarping are still required by preproc yielding the following error: TypeError: preproc() missing 8 required positional arguments: 'delta_te', 'dwell_time', 'manufacturer', 'blip_files', 'blip_enc_dirs', 'unwarp_direction', 'phase_file', and 'magnitude_file'. I now initialize these arguments as None, but if unwarping is optional these arguments should probably be made obstional in pypreclin as well.
• jipdir is first specified, then overwritten with a hardcoded path (/jip/bin)

The code then references an fslconfig python environment that doesn't exist on my system. ($FSLDIR is of course an existing environment variable, but that's not enough).

So, it currently doesn't seem incredibly straightforward to run pypreclin on another system. Perhaps I should look at the singularity image. I haven't used singularity before, but I installed it and figured out how to get into a shell on the image and link some data-folders. If I go to a python3 shell on the singularity image I can import the pypreclin module but if I then want to configure the paths to run preproc I don't know where I should find the jip folder. Your example lists /jip/bin but this doesn't exist. Nor do /etc/jip, /opt/jip, or /bin/jip. For now this is also a dead end...

Are there any superobvious things I'm missing? Have you tested running pypreclin outside of Neurospin or on a 'virgin' system?

Hello,

I'm trying to use pypreclin singularity to do preprocessing of data, and here is my test data

1. Macaque_F99_LR_brian_1p5.nii (template, RAS)
2. Kk_T1_brain_highRes_reorigin.nii (subject T1, IAL)
3. Task.nii.gz (subject task)
   Singularity code used are as follows:

singularity run '/home/sangtian/ST_floder/tmp/pypreclin_test/pypreclin-ubuntu.simg' -V 2 -o 'test_out' -s test -f 'task.nii.gz' -a 'kk_T1_brain_highRes_reorigin.nii.gz' -r 2 -t 'Macaque_F99_LR_brain_1p5.nii' -NA RAS -NF RAS -C /etc/fsl/5.0/fsl.sh -j /i2bm/local/jip-Linux-x86_64/bin

When the program executes to Normalise, an error occurs as:
“Source file 'kk_T1_brain_highRes_reorigin.nii.gz' (LAS) and taget file 'Macaque_F99_LR_brain_1p5.nii' (RAS) must have the same orientation for JIP to work properly.”

I checked the output of code, it had already done Slice timing, reorient, Realign, but empty in the folder Normalization

So then I check the coordinate of subject T1 and template in FSLeyes, the coordinate of subject T1 is original as IAL, and the template is RAS. Then I try to change the coordinate of subject T1 from IAL to RAS, I used the function reorient_image and check_orientation from \pypreclin-master\pypreclin\utils\reorient.py

But even after the preprocessing, the orientation of subject T1 is still IAL, I am confused about it and I am using the following code:

`

from reorient import reorient_image, check_orientation

data_path = '../pypreclin_data/'
T1_data = data_path + 'kk_T1_brain_highRes_reorigin.nii'
RAS_T1_data = data_path + 'RAS_kk_T1_brain_highRes_reorigin.nii.gz'

reorient_image(T1_data, 'RAS', 'RAS_', data_path)
print(check_orientation([T1_data, RAS_T1_data]))

`

the output of the code is: (True, ['IAL', 'IAL'])

From the output, the reorient_image function does not change the coordinate of subject T1, and I don't know what the problem is.

I have also used other packages to orientate the coordinates, e.g. fslreorient2std, fslorient, fslswapdim, and I can convert the coordinates to "LAS" but never to RAS!

At this point, I thought, orient such a process seems to be easily integrated in the singularity step, it seems unnecessary to let us operate separately, so I checked the output folder, and I did find that there was an orient folder in the second step, so I assumed that the container had already completed the reorientation step

But I checked the coordinate of the output file named ‘dRAS_kk_T1_brain_highRes_reorigin.nii.gz’ generated by singularity is still ‘LAS’, so I am really confused about the result, I can't use Pypreclin and FSL's toolkit to solve this coordinate, and also the problem seems to be unsuccessfully handled by the official container, So I really don't know what to do with what's happening right now.

What exactly should I do to keep singularity's program running?

Hello,

I am trying to get Pypreclin up and running in our lab. I am trying just to run the example code on one of my functional and anatomical scans, as a first step (before adding fieldmap distortion correction). The pipeline is getting stuck after getting through most of step 5 (normalization) and throwing an error that traces back to pyconnectome. See the attached screen shot of the error messages. I do have the proper versions of python and all of the other dependencies installed. Do you have any insight? Thanks,

Erin