A Rapid, Accurate and Machine-agnostic Segmentation and Quantification Method for CT-scan-based COVID-19 Diagnostics
Longxi Zhou, et al. "A Rapid, Accurate and Machine-agnostic Segmentation and Quantification Method for CT-based COVID-19 Diagnosis"
Download Data from this Google Drive link. Place the folders datasets/
and checkpoint/
inside the drive in the repository folder.
Run command
conda create -n newenv --file requirements.txt
This will install all the dependencies to run the program. You are free to change newenv
to any name you like for the environment.
Then activate the environment
conda activate newenv
Run command
python test.py
The 3D prediction result will be stored in prediction/
. Viewable png slices of the array will be stored in prediction_visualization/
.
The above is one image slice from prediction_visualization/
. From left to right: (1) original CT image (2) prediction (3) ground truth (4) green regions are true positives, blue regions are false positives and red regions are false negatives.
If you request our training code/simulation model for COVID-19, please contact Prof. Xin Gao at [email protected].