This project tries to preprocess eplilepsy data in a similar way established preprocessing pipelines, like fcon_1000 and C-PAC, would process their data. The preprocessing usually consists of anatomical preprocessing, functional preprocessing, registration, segmentation and nuisance signal removal.

This project has two sub-projects. "pipeline_ext" utilizes nipype's pipeline engine and constructs a more complete workflow; "pure_python" is less complete, but only depends on pure python packages like nilearn, nipy and pypreprocess.

This pipeline depends on external packages like fsl, afni, ants etc through nipype.

pipeline_ext uses nilearn and nipype. Nipype provides an environment that encourages interactive exploration of algorithms from different packages. This eases the design. The packages indirectly referenced here through nipype are ants, fsl, afni, spm.

run as following python pipeline_ext.py -f 109001/func/rest.nii -t template_in_MNI152_2mm.nii.gz --TR 2 -s 109001 --subjects_dir fsdata --slice_times 0 17 1 18 2 19 3 20 4 21 5 22 6 23 7 24 8 25 9 26 10 27 11 28 12 29 13 30 14 31 15 32 16 -o .

This sub-project can apply functional preprocessing and registration on subjects. It is pure Python, meaning no external packages (such as FSL and AFNI) needs to be installed and the code works across platform.

Two packages require direct installation: nipy and pypreprocess. Other packages (such as nilearn and nibabel) will be installed as parts of the package dependencies. Note that this project will only require pure Python parts of pypreprocess, so the instructions on SPM installation can be ignored.

To run "everything" (i.e. to run functional preprocessing and registration sequentially and uninterruptedly), one can type in command line

python pipeline_pure_python.py data_folder

to run with default file locations and settings. To use a more advanced setting, one can add optional arguments -sl --subject_list to point to the "file listing all subjects" (default subject_list.txt), and -ac --anat_preproc_config, -fc --func_preproc_config and -rc --registration_config to provide path to configuration files written as JSON objects. The main function will read these configurations and pass the paramters to function runners.

check here for detailed workflows, docstrings, and more user guides on topics like how to create configuration files, how to create a naming convension, how to use parameters and extra paramters, etc.