A ready-to-use machine learning tool for symmetric multi-modality registration of brain MRI

Preliminaries

Pair-wise registration involves two images: fixed (also known as reference or target) and moving (also know as floating or source). The goal is to find a geometric transform that maximizes the similarity of the fixed image and the deformed moving image. The transform is usually parameterized by a vector of parameters \(\varvec{p}\), and registration posed as an optimization problem: