Important
The package is now available on PyPI: pip install mede
This repository contains the De-Identification of Medical Imaging Data: A Comprehensive Tool for Ensuring Patient Privacy, which enables the user to anonymize a wide variety of medical imaging types, including Magnetic Resonance Imaging (MRI), Computer Tomography (CT), Ultrasound (US), Whole Slide Images (WSI) or MRI raw data (twix).
This tool combines multiple anonymization steps, including metadata deidentification, defacing and skull-stripping while being faster than current state-of-the-art deidentification tools.
You can install the anonymization tool directly via pip or Docker.
Our tool is available via pip. You can install it with the following command:
pip install mede
If you want to use the text removal feature, you also need to install Google's Tesseract OCR engine. You can find the installation instructions for your operating system here. On Ubuntu, you can install it via
sudo apt install tesseract-ocr
sudo apt install libtesseract-dev
On MacOS, you can install it via Homebrew:
brew install tesseract
Alternatively this tool is distributed via docker. You can find the docker images here. The docker image is available for Linux-based (including Mac) amd64 and arm64 platforms.
For the installation and execution of the docker image, you must have Docker installed on your system.
-
Pull the docker image
docker pull morrempe/mede:[tag] (either arm64 or amd64)
-
Run the docker container with attached volume. Your data will be mounted in the
data
folder:docker run --rm -it -v [Path/to/your/data]:/data morrempe/mede:[tag]
-
Run the script with the corresponding cli parameter, e.g.:
mede-deidentify [your flags]
De-Identification CLI
usage: mede-deidentify [-h] [-v | --verbose | --no-verbose] [-t | --text-removal | --no-text-removal] [-i INPUT]
[-o OUTPUT] [--gpu GPU] [-s | --skull_strip | --no-skull_strip] [-de | --deface | --no-deface]
[-tw | --twix | --no-twix] [-p PROCESSES]
[-d {basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} [{basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} ...]]
options:
-h, --help show this help message and exit
-v, --verbose, --no-verbose
-t, --text-removal, --no-text-removal
-i INPUT, --input INPUT
Path to the input data.
-o OUTPUT, --output OUTPUT
Path to save the output data.
--gpu GPU GPU device number. (default 0)
-s, --skull_strip, --no-skull_strip
-de, --deface, --no-deface
-tw, --twix, --no-twix
-w, --wsi, --no-wsi
-p PROCESSES, --processes PROCESSES
Number of processes to use for multiprocessing.
-d {basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} [{basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} ...], --deidentification-profile {basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} [{basicProfile,cleanDescOpt,cleanGraphOpt,cleanStructContOpt,rtnDevIdOpt,rtnInstIdOpt,rtnLongFullDatesOpt,rtnLongModifDatesOpt,rtnPatCharsOpt,rtnSafePrivOpt,rtnUIDsOpt} ...]
Which DICOM deidentification profile(s) to apply. (default None)
If you use our tool in your work, please cite us with the following BibTeX entry.
@article{rempe2025identification,
title={De-identification of medical imaging data: a comprehensive tool for ensuring patient privacy},
author={Rempe, Moritz and Heine, Lukas and Seibold, Constantin and H{\"o}rst, Fabian and Kleesiek, Jens},
journal={European Radiology},
pages={1--10},
year={2025},
publisher={Springer}
}