pykao / modified-3d-unet-pytorch

This repository implements pytorch version of the modifed 3D U-Net from Fabian Isensee et al. participating in BraTS2017

Modified-3D-UNet-Pytorch

This repository implements the modified 3D UNet architecture in pytorch from Brain Tumor Segmentation and Radiomics Survival Prediction: Contribution to the BRATS 2017 Challenge Fabian Isensee et al. participating in BraTS2017.

You are able to find the PyTorch version of the modified 3D UNet in model.py

The model works but you have to work on data loading part and training part.

Please refer to our BraTS2018-tumor-segmentation repo if you would like to use this model on brain tumor segmentation.

modified-3d-unet-pytorch's Issues

demo to run 3Dunet

hello, thanks for you work, would you mind to share you demo file to run this model? Appreciate

training part and data loading part

Thanks for the great respo.
I'm really enjoying your work, but I'm a starter in pytorch.
Would it be possible for you to release the training part and data loading part? So this project can be used to training models based on local data.

Looking forward to your reply.
Edward

BraTS2018-tumor-segmentation repo link

Hi @pykao
Thanks for great repo. I am just wondering if you can provide proper link for BraTS2018-tumor-segmentation as you mentioned at the end of the README.md-

Please refer to our BraTS2018-tumor-segmentation repo if you would like to use this model on brain tumor segmentation.

Best,
Azam.

about data process

Hello,

Thank you very much for your work. I have a question about how to process data. In what form is the data used as input to the network? Or can you open some data?

Thanks in advance!

input shape of the modified unet

Hi

Thanks for sharing your work,

Does the modified unet expect a fixed shape like (1,1,128,128,128) or will any shape do ?

Thanks in advance.

unable to figure out training script

for epoch in range(max_epochs):
    # Training
    for orig, segmented in training_generator:
        # Transfer to GPU
        orig, segmented = local_batch.to(device), local_labels.to(device)

orig and segmented are torch tensors that represent the images that i need UNet to use to train. How do I implement the training script for this? I'm a bit confused.

About the “id_name_conversion”

Thanks for your code and I have a problem.
When I run the N4ITKforBRATS2017.py，I got an error :/traindata/id name conversion.txt not found. What is the content of this txt file?