In this folder, we perform an analysis of the dataset focusing on the lesion segmentations based on M0 manual segmentation and M12 predicted segmentations. In more detail, it looks at the correlation between age, phenotypes, lesion counts, lesion volume, lesion length, and lesion distribution. The steps are :
- Set up the environment
- Generate the dataframe
- Analyse the dataframe
To install the required libraries run :
pip install -r requirements_dataframe.txt
It is also required to install SpinalCordToolbox 6.0 :
Installation link : https://spinalcordtoolbox.com/user_section/installation.html
To generate the dataframe run :
python generate_dataframe.py --data /path/to/CanProCo --lesion /path/to/lesion/segmentation --discs /path/to/discs/segmentation --spinal-cord /path/to/spinal/cord/segmentation --timepoint M0 --exclude-file /path/to/exclude/file --output /path/to/output/folder
Note It uses the file
image.pyto fix the orientation of masks if they are not the same as the image's orientation
The following Notebook dataframe_analysis.ipynb details the analysis of the generated dataframe. To use it, update the link to the CSV dataframes.