I have a large dataset of head CTs I would like to normalize :
1/ Same size 512x512x40
2/ Same RAI Orientation
3/ Same centering
4/ Same head alignment
I have a head CT template that I could use to center and rotate my raw CTs, but not to resample because depth size is way too much (180) compared to my data (40). (registering to this template implies resampling to the same dimensions right ?)
I need some guidance on the steps I should follow, here is my plan :
To solve 2/ : read image with sitk, SetDirection to RAI space
To solve 3/ and 4/ : I don’t see how I can use the template to align and center without having to resample the registered image to the template size, this is my first question.
Case A : My current guess is that I should manually rotate and translate a CT, resample it to 512x512x40 and use it as my fixed image ?
Is it really clean or is another solution favored ?
Case B: if I could possibly learn a rigid transform to translate and rotate will keeping the original dimension.
To solve 1/ : would be solved by resampling manually if case B. If case A, image is already at correct size.
This all seems pretty straightforward, but more difficult without practicing. There might be a simpler solution for all this, if so please tell me ^^