JuliaWolleb

This is a left-over of the original diffusion model implementation we built our method on (https://github.com/openai/guided-diffusion). There you need to specify the image dimesion to be in (64,128,256,512). If you...

Hi Sorry for this confusion. In the original BRATS dataset, the images are of size (240,240). When you then compute `image = image[..., 8:-8, 8:-8]`, you crop them to a...

Hi Sure, you need to have a one-hot-encoding for the multiple classes. So you will have one binary output channel per class. The only thing you need to change is...

Hi For our application of anomaly detection, during testing, we are interested in generating only healthy subjects. So any input image (both healthy and diseased) are passed through the generator...

Sure! 1. During training, the difference map is not used. The difference map is only used for evaluation. This "if" statement says that during evaluation, we take the difference between...

Hi Well, you need to provide the error message so that I can help you. The shapes seem to be correct. k should indicate the class label, so it should...

Hi There seems to be a problem with apex. Did you install it correctly? You could try to set the optimization level to 'O0' instead of 'O1', or leave out...

You can choose the optimization level 'O0' in lines 132 and 133 in the file _model/DeScarGAN.py_, as well as in lines 77 and 78 of the file _Evaluation/Evaluation_Synthetic_Dataset.py_ and in...

Hi Thanks for your message. I will check it out whether I get the same error message.