I am working on MR images to synthetic CT and my results are not very satisfactory based on the output. I am new to pytorch so I don’t have experience in it.
Now, I am planning to merge two models.
1st model that I was using to do my research takes the input MR and CT images to train the network and once trained it provide me the synthetic CT. Model is trained in a way that it should not loose the HU values while reconstructing the images because HU is the most important part for the medical CT images.
Now I am planning to implement super resolution over this network.
I would like to hear the suggestion from the experts that should I take the output of my network to the input of super resolution model to get much improved result or there can be some other better solution.