STIR Project Files

Python scripts for creating, parsing, and viewing MRI databases consisting of both MINC and Dicom imaging formats.

This combination of scripts is designed to create an organized database of scans that can be analyzed, registered, and segmented in an automated fashion.

Contents

pyscripts - Python scripts for database maintenance and applying conversions across the entire database.

bin - Bash script wrappers for the registration Python scripts. Add to your PATH environment variable to make execution a lot easier.

analysis - Python scripts for image segmentation/analysis as well as registration quality scoring. Example files included.

examples - Example Bash scripts that demonstrate the usage of these software tools on an example database.

Dependencies

Python libraries used in these scripts include:

• [PyDicom][pydicom]
• [libmagic][libmagic]
• [OpenCV (cv2)][opencv]

Command line tools called as subprocess commands:

• [Minc Toolkit][minctool]
• [mri_convert (from Freesurfer package)][mriconvert]
• [dcm2nii][dcm2nii]
• [nifti_reg][niftireg_link]

Due to some issues with subprocess not using the correct path, the absolute path to each command is specified in the python code. Replace this with the absolute path for your system (output of the command which <command>).

Programs

• createdb <name_of_output.csv> </path/to/database>

  • Creates a CSV file with all scans in the database
  • Assumes a file structure of patient/date/scan_modality/scanfile.dcm

• searchdb <id/modality/list> <search term> </path/to/database.csv>

  • Searches for a scan in the generated CSV file by id or modality.
  • Returns path of all scans that match the search parameters.
  • Alternate parameter list lists all imaging modalities in the database.

• db2nii <source_dir> <output_dir>

  • Given a source database of DICOM and MINC files, creates a new database with scans converted to NIFTI format.
  • Most registration software requires that scans be in the NIFTI format.
  • Assumes a file structure of patient/date/scan_modality/scanfile.(dcm/mnc)

• db_reg <source_dir> <output_dir> <optional: # of cores (default 8)>

  • Given a source database of NIFTI files, register a patient's diffusion scan to their structural T2 scan.
  • Registration is performed intra-patient only.
  • Assumes a file structure of patient/date/scan_modality/scanfile.nii.gz
  • Warning: this code, by default, overwrites the original input scan with the registered scan. Make a copy of your database before running this program.

• db_cross_reg <source_dir> <output_dir> <reference_T2_path> <optional: # of cores (default 8)>

  • Given a source database of NIFTI files, register a patient's structural T2 scan to a reference T2 scan.
  • Registration is performed inter-patient only.
  • Assumes a file structure of patient/date/scan_modality/scanfile.nii.gz
  • Warning: this code, by default, overwrites the original input scan with the registered scan. Make a copy of your database before running this program.

Lonestar 5 Notes

When running on Lonestar 5, some dependencies are pre-installed. Add these programs to your path:

• mri_convert: /corral-repl/utexas/poldracklab/software_lonestar/freesurfer/bin/mri_convert
• dcm2nii: /corral-repl/utexas/poldracklab/software_lonestar/local/bin/dcm2nii
• reg_aladin (from the nifti_reg package): /corral-repl/tacc/bio/STIR-data/rstevens/nifti_reg/bin [pydicom]: https://github.com/darcymason/pydicom [libmagic]: https://pypi.python.org/pypi/python-magic [opencv]: http://opencv.org/ [minctool]: https://github.com/BIC-MNI/minc-toolkit [mriconvert]: http://freesurfer.net/ [dcm2nii]: https://www.nitrc.org/projects/dcm2nii/ [niftireg_link]: http://cmictig.cs.ucl.ac.uk/wiki/index.php/NiftyReg