Objects change their form over time.
Undoing these changes makes the two images of the same object more comparable.
This process of finding a transformation of one of the images, such that it gets as similar as possible to the other one is called deformable registration.
Deformable registration can lead to improvement of many medical procedures, both diagnostic and interventional.
The main reason is that deformations of the human body are present in many settings and for many imaging techniques.
Always, when these deformations occur between different scans, there is potential for the current methods to be improved by using deformable registration techniques.
Example applications include follow-up studies, image fusion from different modalities and treatment planning.
The focus of our work is two-fold: we try to understand and improve the existing registration methods and to solve clinical problems by applying the developed methods.