3D–2D registration of vascular structures
Timur Aksoy, Gozde Deniz Unal, Franjo Pernuš, Žiga Špiclin · 2019
In this chapter, a mathematical definition of the 3D–2D registration problem is given and 3D–2D methods are classified with respect to five criteria: type of image modality, spatial transformation, dimensional correspondence, number of 2D views and registration basis. A detailed review of registration bases that define how similarity is assessed between the images is performed, followed by a detailed classification and review of techniques for estimating the spatial transformation via optimization of the similarity. The validation procedures presented constitute an important phase of the development of the registration method and are used to determine the potential of a method for clinical application. This chapter concludes by showcasing the results of state-of-the-art methods on clinical data and discussion of the open challenges in the translation of 3D–2D registration technology into clinical applications.