Introuction

• This is a pytorch implementation of part of 3DGAUnet

Prerequisites

• Python >= 3.7.9
• Pytorch >= 1.6.0
• tensorboardX >= 2.1
• matplotlib >= 2.1

Pipeline

Data

Volumetric data in .mat format should be placed in'/src/volumetric_data/', and change the directory in 'paramt.py' accordingly.

Training

cd src run python main.py on GPU or CPU. Of course, you need a GPU for training until you get good results. I used one GeForce RTX 4090 in my experiments on 3D models with a resolution of 64x64x64.

• Model weights and some 3D reconstruction images would be logged to the outputs folders, respectively, for every model_save_step number of step in paramt.py. You can play with all parameters in paramt.py.

Generation of synthesis data

To generate volumetric data from the trained model, you can run python main.py --test=True to call tester.py.

Pre-trained Model

Pretrained models are in the outputs folder. Then run python main.py --test=True --model_name=pancreas_pretrained. You will find the outputs in the test_outputs folder within pancreas_pretrained.