Comparison of distance measures in Multimodal registration of medical images

Abstract

In this thesis I have worked with medical image registration, in particular registration of multimodal images. I have tested the image registration software package FAIR developed by Jan Modersitzki, with particular focus on the choice of distance measure in image registration. I have tested the distance measures included in FAIR, in particular Normalized Gradient Field, the distance measure developed by Modersizki, and compared them. In addition, I have developed a new distance measure called Normalized Hessian Fields and implemented it with the FAIR software. This has been compared to the other distance measures. I have also tested a combination of Normalized Gradient and Hessian Fields. The registration has been performed on multimodal brain data and kidney data provided by Haukeland University Hospital, Bergen.