Researchers: Angelica Aviles-Rivero, Carola-Bibiane Schönlieb, Keith Goatman (Toshiba Medical Systems)
Image registration is a common process for medical imaging, in which given an image sequence the main goal is to find a set of transformation that find spatial correspondences. This task is commonly set an optimization process, which demands high computational time due to the resulted high-dimensional parametric problem. A feasible and attractive option to achieve computational tractability and speed up the solution is by learning the parameters, and in particular, by using sophisticated algorithms derived from deep learning. Thus, the aim of this work is to propose a new solution that can learn effectively the image registration parameters.