Medical image segmentation pipeline to segment background and three lesions:

• Ground-glass opacity
• Consilidation
• Pleural Effusion

[]

• Ubuntu > 18.04
• Python > 3.6
• NVIDIA card with CUDA support (RTX 1070Ti or better)

Download the code repository via git clone to your disk. Afterwards, install all required dependencies, download the dataset and setup the file structure.

git clone https://github.com/tyommik/covid19_ct_lesion_seg
cd covid19_ct_lesion_seg

conda env create --file environment.yml
jupyter lab

Run train.ipynb using your own data

Using private dataset. There is no option to provide it :(

<demo> : http:https://covid19.shibaev.info