We adopt some code from https://github.com/milesial/Pytorch-UNet
torch==1.8.1+cu111
matplotlib==3.4.2
pandas==1.2.4
torchnet==0.0.4
scipy==1.6.3
numpy==1.19.5
scikit_learn==1.0
SimpleITK==2.1.1
tensorboardX==2.4
- Download ModMedData CT dataset https://mosmed.ai/datasets/covid19_1110/
- Config
base_path = './'
in preprocess.py
3. run python preprocess.py
1.Config train.py
basepath = './'
train_epoches = 500
val_period = 5
2.run python train.py
You can download our trained weight file here https://drive.google.com/file/d/1GnOtJ4d4CE1vfiG15FYYFgmtr40y-2Dw/view?usp=sharing
- Config
Analysis.py
basepath = './'
model_path = 'model/405.model'
- run
python Analysis.py
1.Config test.py
model_path = 'model/405.model'
test_image_path = 'studies/CT-0/study_0200.nii.gz'
threshold = [0, 3299.032990329903, 47378.47378473785, 86080.86080860808, 214349]
2.run python test.py