For the clinical diagnosis of myocardial perfusion images there is a need to register 3D datasets originating from different modalities. CT datasets provide good resolution of anatomic structures and are needed for attenuation and scatter correction during the process of SPECT image reconstruction. It is mandatory for this process to have CT and SPECT 3D volumes realigned properly so that reconstruction algorithms can provide physicians with valuable data. Several rigid registration techniques are implemented in commercial solutions for hospitals. These techniques cannot take into account any non-rigid deformations, which may occur e.g. because of respiratory motion, the beating of the heart or shifts of the patient.
This book shows the use of integrating a non-rigid registration approach into the process of reconstructing SPECT datasets of the myocardium. The results of the optimization of this process, as well as its application to clinical images are discussed. Besides an overview of related works and patents in the field of non-rigid image registration, this book presents the theoretic background needed to understand the concepts and methods behind this kind of registration.