DICOM Utils

Collection of helpful scripts and Python methods for working with DICOMs.

Setup

This repo can be installed with pip. To install to a virtual environment:

1. Run make init to create a virtual environment with dicom-utils installed.
2. Call utilities with venv/bin/python -m dicom_utils

Alternatively, install the repo without a virtual environment and run the entrypoints provided by setup.py

1. pip install . or pip install -e .
2. Run utilities anywhere as dicomcat, dicomfind, etc.

Usage

The following scripts are provided:

• dicomcat - Print DICOM metadata output as text or JSON
• dicomfind - Find valid DICOM files, with options to filter by image type
• dicomphi - Find and overwrite PHI across DICOM files
• dicom2img - Convert DICOM to static image or GIF
• dicom_types - Print unique values of the "Image Type" field
• dicom_overlap - Find StudyInstanceUID values shared by files in two directories

PHI Anonymization Rules

If anonymization is enabled when running dicomphi, fields defined in this script are anonymized with the exception of fields which are affected by additional rules located here.

pynvjpeg

To install pynvjpeg for accelerated JPEG2000 decoding, install the j2k extra.

$ pip install -e ".[j2k,dev]"

The following steps may be required:

1. apt install cmake

2. Add CUDA tools to path

CUDA_FOLDER="cuda-12.0"
export PATH="/usr/local/$CUDA_FOLDER/bin:$PATH"
export LD_LIBRARY_PATH="/usr/local/$CUDA_FOLDER/lib64:$LD_LIBRARY_PATH"

3. apt -y install python3-pybind11