Quickstart
Prerequisites
Note
Optical Character Recognition (OCR) is an intensive resource-consuming process and it is recommended to have at least 4 GB of free RAM for running kskit functionalities related to image deidentification.
Info
kskit relies on easyOCR
which uses PyTorch
. If your system does not have a GPU, consider installing PyTorch
CPU-only version (more information here).
- For installing
PyTorch
andPyTorch Vision
CPU-only dependencies on a Linux system, run the command below:
- If you have a GPU on your system, please install the correct version of
PyTorch
andPyTorch Vision
for your computer (PyTorch Installation Guide).
Installation
For installing kskit
tools, run the following command:
After that, you'll have to define an environment variable called DP_HOME
. This
variable is used to locate your data directory where you'll:
- put your DICOM files for submitting them to the deidentifier tool
- find the output of the deidentifier tool (deidentified information)
- define referentials used by the package (
recipe.json
,ocr_deid_words.txt
)
To define this data directory, run the following command:
Info
You can set this folder wherever you want. However, keep in mind that this folder is essential for the package and you'll probably open it more than once. So, don't put it somewhere too complicated to access.