Dominik Müller

Hey @LHJ97, I apologize for the late response. Mhm, you are right. Looks like the number of patches is zero. It could be that somehow empty batches are generated due...

Hey @aleemsidra, it seems to be that you try to run 2D code with 3D data? Please provide more information to debug like a full error log or a reproducible...

Hey @aleemsidra, sorry for the late response. Could you please also add another screenshot of the Preprocessor? And is this the KiTS19 Notebook from the examples? Cheers, Dominik

Hey @Jugaris89, > what is the need of the "Find out good voxel spacing" part? I see, the "x", "y" and "z" calculated there are not used in the rest...

Hey @Jugaris89, sorry for the late response. > In the visualization of the predicted classes (utils/visualizer.py) you consider only 2 classes to visualize: imagen I understand this is because in...

I added the absolute transformation of voxel spacing in the DICOM IO interface. Could you update to the latest MIScnn version 1.0.4 and try it again. ```sh pip install miscnn...

No problem. > What dice score do you get for the pancreas? --> It starts by 0.3... and decreases until

Hey @Jugaris89, sorry for the late reply again... How is it going with your pancreas segmentation? Did you make progress? Have you e.g. tried further increasing the maximum clipping range...

Hey @Jugaris89, mhm, could you please provide the full error log for this issue using the updated miscnn version? As well as your code how you applied the resampling? Cheers,...

Hey @bhralzz, MIScnn utilizes Tensorflow for Neural Network actions as well as their multi-gpu API. Sadly I'm unfamiliar with this error, however Tensorflow has various issues on multi-gpu support with...