Digitiz

an OMR tool that automates data extraction from paper medical records

Dependencies

Python 3
Python Imaging Library (PIL) (python3-pil, distributed as part of python3-Pillow)
Python lxml (python3-lxml)
NumPy (python3-numpy)
SciPy (python3-scipy)
scikit-image (scikit-image)
opencv (opencv-contrib-python)
onnxruntime (onnxruntime)
pathlib (pathlib)
flask (flask)

All of these can be installed used pip or conda. If you are planning to plug in an external UVC-enabled USB camera to use the live alignment feature, make sure to download opencv-contrib-python, since the package hosted under opencv-contrib does not provide video device support via OpenCV.

Getting Started

Start the server using the following line:

python3 backend/server.py

Running Tests

To run a test suite, first navigate to backend/tests and run the test_end_to_end.py script on the desired test set. For example, to run PIH Admission test set execute the following commands from the repository root:

cd backend/tests/
python3 test_end_to_end.py --INPUT_FILE test_sets/pih/admission.csv

Generating New Tests

To generate new test cases from each scanned record run the server in debug mode, making sure to specify your desired camera and output directory. For example:

python3 backend/server.py --save-debug --camera_index 1 --upload_folder [upload_folder]

...where [upload_folder] is the absolute path to the directory into which the server should write all new test files.

Find Out More