I have a pair of pretreatment CT and posttreatment CT. I am especially interested in the registration accuracy close to the tumor region in the lungs.
My previous algorithm is based on extracting lungs and registering only extracted lung CTs with a multi-resolution approach. I want to improve registration close to the tumor region. I am planning to apply a mask in those regions to force the algorithm to focus on ROI.
Is there a way to compose different parameter maps with different masking conditions? Because what I understood now is that if I set a mask, it is implemented for all parameter maps.
To illustrate I want transformations in the following order: Rigid (without mask) => BSpline (without mask) => BSpline (with mask on ROI)