valabregue

Hello I came also across this strange result. And to investigate I change the code so that instead of returning the mean over the volume it return the volume values...

I dig a little bit more, and get this : So I start witht a basic Rigid coregistration with ```python elastixImageFilter = sitk.ElastixImageFilter() elastixImageFilter.SetFixedImage(sitk.ReadImage("fixedImage.nii")) elastixImageFilter.SetMovingImage(sitk.ReadImage("movingImage.nii")) elastixImageFilter.SetParameterMap(sitk.GetDefaultParameterMap("rigid")) elastixImageFilter.Execute() ``` ``` sitk.PrintParameterMap(elastixImageFilter.GetTransformParameterMap()...

Ok I found almost the solution: if one do a affine registration (thanks to @kaspermarstal https://github.com/SuperElastix/SimpleElastix/issues/91#issuecomment-300691102 I thing it works the same but with a `EulerTransform` we get after a...

Great ! this will be a nice improvement The thing with gpu support, is that it will not be available in pytorch DataLoader since they are cpu only . (correct...

> maye there is no point in doing this. no, I am very confident that this will be useful. may be the DataLoader strategy will change in a near future,...

ok, If I understand correctly, you say, one should put num_worker=0, to stay in the main process, ... that make sens, especially if one have only one GPU, to avoid...

Hi just out of curiosity, @Spenhouet how to you manage to match the output of the model (I guess same size as the T1 input) with the label that have...

interesting way to spare space with multiple label file covering only a subpart of the total FOV ... I also struggle to get enough memory with 3D cnn and 15...

Hi thanks for the proposition, this is indeed a very important topic, since in practice it is often that instead of using a super GPU card 90% of the time,...

thanks for the detail, I think I now get a better understanding ok, I see now the difference between 'basic' and 'merge incomplete' It seems to me a quite small...