I want to preprocess MRI t1 for PARKINSON's disease, in these steps I used registration with mni152, but the dimensions changed, I think I lose some information in this step, (eg my images are 256256 192 in the mni152 images are 128182128), also after the processing steps I want to send these images to the deep network and in this step I also need to change the dimension, these two resizing steps will destroy Loss of information and decrease in accuracy is not recognized?
what should i do, is it necessary to use registration?