DICAT is a simple graphical tool that facilitates DICOM (Digital Imaging and Communications in Medicine) de-identification directly on a local workstation. It was designed to work on all major operating systems (Windows, Linux and OSX) and is very light in terms of dependencies (Python). Binaries (with no dependencies) have been compiled for all operating systems and made available with each release of DICAT.
With the increasing use of web-based database systems, such as LORIS (Das et al., 2011, Das et al., 2016), for large scale imaging studies, de-identification of DICOM datasets becomes a requirement before they can be uploaded in such databases.
Typical Flow Chart of DICOM de-identification.
Before DICOM datasets can be uploaded into a web-based database, identifying information stored in the DICOM header (such as patient name, date of birth) should be removed.
DICAT produces two archival outputs: a back-up of the original DICOM files, and a de-identified DICOM dataset that can then be uploaded or transferred to other systems.
DICAT also features an ID key log that can be used to keep a record of the original candidate name (participant/patient) linked to their anonymized study identifier, for reference by study coordinators.
DICAT was first developed during the 2014 and 2015 brainhacks held at the Organization of Human Brain Mapping (OHBM) conferences.
Installation instructions vary depending on the operating system used. See below for detailed information.
Running DICAT will open a window with three different tabs:
- A simple "Welcome to DICAT" tab giving a short description of the tool
- A "DICOM de-identifier" tab, in which DICOM de-identification will take place
- An "ID Key" tab, containing the key between candidates' name and their IDs
Welcome page of DICAT.
Executables of DICAT have been created for most systems and can be found with each release of DICAT in the Github repository. Download the executable relevant to your system and move it to any folder of your choice.
To open DICAT, simply double click on the executable.
Please note that for some obscure reason, DICAT is extremely slow to run on Windows OS.
Before running DICAT, make sure your system contains a Python 3 compiler with the TkInter library (usually, TkInter comes by default with most Python installations).
For Ubuntu distributions, TkInter can be installed via apt-get:
sudo apt-get install python-tk
The PyDICOM package is also required by DICAT.
For most platform, PyDICOM can be installed via pip:
pip install pydicom
To install DICAT source code on a computer, download and save the content of the current Github repository into a workstation.
DICAT can be started by executing DICAT.py script with a Python compiler.
On UNIX computers (Linux and Mac OS X), open a terminal, go to the main
directory of DICAT source code (dicat directory) and run the following:
python DICAT.py
DICOM de-identification with the DICAT GUI
In the "DICOM de-identifier" tab (1), use the select button (2) to choose a directory containing DICOM files to de-identify.
Once a directory containing DICOM files have been selected (as described in the above section), the DICOM fields can be viewed when clicking on the “View DICOM fields” button (3).
The DICOM identifiable fields will be displayed in a table with editable fields (4).
Users can choose to delete all identifiable fields using the “Clear All Fields” button (5).
Users can also directly edit the fields (6) in the table and all values present in the table will be inserted into the corresponding DICOM fields in the imaging files.
Note that the PatientName field is required (7) and will need to be filled with new IDs in order to label the scan for that session.
Finally, once the user has finalized the edits, clicking on the “De-identify” button (8) will run the de-identification tool on the DICOM dataset.
Mass DICOM de-identification using mass_deidentify.py
The script mass_deidentify.py in the dicat directory allows a user to
mass de-identify a series of DICOM studies. Running mass_deidentify.py -h
will display the following information on how to run the script:
The ID Key feature of DICAT allows storage of the key between identifiable candidates's information (Real Name and Date of Birth) and its study’s identifier. This information will be stored locally on the workstation within an XML file (candidate.xml) in DICAT's directory. See the following figure for detailed information on how to use this feature.
ID key feature of DICAT.
This feature (1) allows storage of the mapping information between candidates’s information and study IDs. This information will be stored in an XML file that can be either created (2) or opened (3). Changes will be automatically saved.
A candidate (participant/patient) can be looked up using the “Search candidate” button (5) after having entered either the “Identifier” or the “Real Name” text fields available in (4).
The “Clear fields” button (6) allows clearing the text in those text fields.
A new candidate can be registered using the “Add candidate” button (7) after having entered the “Identifier”, “Real Name” and “Date of birth” information in the text fields of (4).
Clicking on a subject row (8) of the table displayed at the bottom of the application will automatically populate the text fields (4) with the information of the candidate.
The “Real Name” or “Date of birth” of that candidate can be edited if needed by altering the field and clicking on the “Edit candidate” button (9).
Finally, the data table of candidate is sortable by clicking on any of the column headers (10).
Ayan Sengupta uam111@gmail.com - Concept, Pydicom implementation
Cecile Madjar cecile.madjar@gmail.com - GUI implementation, PyDICOM implementation, python integration of DICOM-toolkit, ID key
Dave MacFarlane david.macfarlane2@mcgill.ca - ID Key
Samir Das samir.das@mcgill.ca - Concept and guidance
Daniel Krötz d.kroetz@fz-juelich.de - Documentation, testing on Windows
Christine Rogers christine.rogers@mcgill.ca - Documentation
Leigh Evans evansleigh26@gmail.com - Video tutorial
Derek Lo derek.lo@mcgill.ca - Logo design