MR_preproc_dash

Dashboard to monitor MR image preproc pipeline status

Basic workflow

• run preproc status script anytime after preproc job has been submitted

• status script

  • input:

    • data_path to preproc dir containing all the subject subdirectories
    • save_path to dump summary csv and dataframe pickle

  • output: status summary as a csv and a dataframe (or color coded dataframe and pairplots in notebook)

• functionality

  • checks preproc output directory tree for status "exists" or "missing". Note: this is a squential status overwritting process; i.e. "file missing" implies that the subject, timepoint and output directories exists but a particular file is missing.

    • timepoint dirs (per subject)
    • MR output dirs (per timepoint)
    • MR mnc files (per output dir)

  • extracts registration parameters by inverting registered image (stx, stx2) for each subject and looks for intra-subject (timepoints) outliers.

Code structure

• ./lib : useful defs
• ./notebook/MR_preproc_dash_test_code.ipynb: test-run notebook (This will produce color-coded dataframes!!)
• ./notebook/MR_preproc_dash_registration_metrics.ipynb: notebook for creating reg_param outlier pair-plots
• ./run_test.py command-line code (This will dump summary data to a csv and df pickle)

Examples

- source /ipl/quarantine/experimental/2013-02-15/init.sh (at BIC) 
- run_test.py:  python run_test.py --data_dir /data/ipl/scratch03/nikhil/MR_preproc_dash/mahsa_preproc_test_data/ --save_path /data/ipl/scratch03/nikhil/MR_preproc_dash/preproc_dash

Limitations

• Does not QC or check contents of the file during directory tree search

prerequisites

• python 2.7+ (3 preferred)
• pandas
• seaborn (for pairplot visuals)
• https://github.com/vfonov/nist_mni_pipelines (needs iplPatient.py in PYTHONPATH)