Semi-supervised-Echocardiography-Segmentation

基于对抗网络的三维超声心动图左心室心肌半监督分割方法

Envirment require:

1. Pytorch=1.4.1
2. visdom
3. Python3.6

Training:

1. python -m visdom.server
2. python tran.py --data_type==0 --loss_type=0 --resume_epoches=0 --vis_name='GAN_Seg' --data_type: Synthetic data(0), Animal data(1) --loss_type: Supervised(0), Semi-supervised(1), Semi-supervised+Self-supervised(2) --resume_epoches: Previous training *pth --vis_name: visdom view name

Testing:

1. python -m visdom.server
2. python test.py --data_type==0 --loss_type=0 --resume_epoches=0 --vis_name='GAN_Test_Seg' --data_type: Synthetic data(0), Animal data(1) --loss_type: Supervised(0), Semi-supervised(1), Semi-supervised+Self-supervised(2) --resume_epoches: Previous training *pth --vis_name: visdom view name

During training or testing, open localhost:8097 to see the visulized variable.