Introduction

DEIDENTIFICATION of dicom studies, for participating to MIDI-B De-identification challenge (https://www.synapse.org/Synapse:syn53065760/wiki/627876)

installation

Clone the repository
Create an enviroment python -m venv venv
Activate the venv venv\Scripts\activate
Install all pakages pip install -r requirements.txt
Install keras-ocr pip install -q keras-ocr

lets run:

into debug_de_identification.py insert the input parameters: - input_data_folder = r'C:\challenge_testdata\input_data' - rules_file = './custom_rules.json' (here already the rules used for the challenge) - basic_profile_file = './base_anonymization_profile.csv'
python debug_de_identification

Outputs

in the input data folder have been created the data and the mappings folder
a log file has already created.

Contribute

https://dicom.nema.org/dicom/2013/output/chtml/part15/sect_E.3.html#sect_E.3.1

package used

pydicom==2.4.4
numpy
matplotlib
scikit-image
pillow
tensorflow==2.15
keras-ocr
thefuzz