valemto/nrrdify

utility scripts for converting dicom to nrrd (DWI)

Warning: these scripts have not been widely tested and should only be used in research.

These scripts convert directories of DICOM data into nrrd files using 3D Slicer's DWIConvert utility.

• convertall.py - runs DWIConvert on all series folders in the hard coded directory structure

  • input is dicom directory (currently hard coded)
  • output is directory named "converted" with the same patient/study naming convention but with each study directory containing a series file as a .nrrd along with the output stdout and stderr of the converter command for diagnostic purposes.
  • a convert.log.txt at the top level of the converted directory is kept for overall diagnostics
  • NOTE outuput of this is still PHI

• renameNRRDs.py - removes patient ID

  • input is a directory output from convertall.py
  • output is a directory of the form: `converted-rename/patient-#####/patient-#####-study-#####/patient-#####-study-#####-series-##.nrrd (No identifiers, just images)
  • creates a patientMap.json that maps MRN to patient-##### (contains PHI)
  • creates a studyMap.json that maps MRN and original study subdirectory name to study-##### (contains PHI)

• studyMeta.sh creates tables of patient ID information (optional - an example of how to extract data at the study level)

  • in the source data directory (hardcoded) creates the following:
    • studyMeta.dump with a line for StudyData,PatientID,PatientAge for each study (contains PHI)
    • studyMeta.txt with same data on one line (contains PHI)

Background: These scripts were developed to process dicom directories in the format exported from the Partners mi2b2 system and would need to be customized for any other use.

Usage: These have been used on a mac, but probably also work unchanged on linux, but not on windows unless using a linux compatibility layer.

TODO: These script could be useful for other purposes and should be generalized at some point to handle different directory layouts and command line options.

Sponsored by: MI2B2 ENABLED PEDIATRIC RADIOLOGICAL DECISION SUPPORT, NIH grant R01EB014947