This package contains a python program that demonstrates automated C2-C7 Cobb angle measuring. The program is tested on Windows 10 and Ubuntu 20.04.
Unlike the web application and executables, this python version uses GPU when it is available.
Python 3 is required to run the program.
The main dependencies of the program are the following python packages.
torch==1.10.1
numpy==1.21.0
scipy==1.7.3
pydicom==2.2.2
segmentation-models-pytorch==0.2.1
The software may or may not work with other package versions.
Full list of the dependencies is available in requirements.txt
.
It takes about five minutes to install the dependencies.
Replace python
and pip
commands with python3
and pip3
respectively if needed.
Create an environment.
python -m venv c2c7
Activate the environment.
(Windows)
c2c7\Scripts\activate
(Linux)
source c2c7/bin/activate
pip install -r requirements.txt
Download c2c7_ENetB4.pth
from google drive and save it as weights/c2c7_ENetB4.pth
.
Run the program in a terminal as follows.
python c2c7_cobb.py <input> <output>
, where <input>
is the input x-ray image and <output>
is the output image.
<input>
image can be a DICOM image whereas <output>
cannnot.