Medical Diffusion: This repository contains the code to our paper Medical Diffusion: Denoising Diffusion Probabilistic Models for 3D Medical Image Synthesis
According to the paper, the MRI images are histogram-based normalized and further standardized to -1 and 1 using min-max scaler. My question: how did you choose the min and max values for scaling? Were these values calculated per volume or per dataset? And after inference, do you scale back the intensity values or just keep them in the range of -1 and 1?
According to the paper, the MRI images are histogram-based normalized and further standardized to -1 and 1 using min-max scaler. My question: how did you choose the min and max values for scaling? Were these values calculated per volume or per dataset? And after inference, do you scale back the intensity values or just keep them in the range of -1 and 1?
Thank you!