Björn Eskofier

Non-rigid 3D Spect/ct Image Registration: Registration of Myocardial Datasets Björn Eskofier

For the clinical diagnosis of myocardial perfusionimages there is a need to register 3D datasetsoriginating from different modalities. CT datasetsprovide good resolution of anatomic structures andare needed for attenuation and scattercorrection during the process of SPECT imagereconstruction. It is mandatory for this process tohave CT and SPECT 3D volumes realigned properly sothat reconstruction algorithms can provide physicianswith valuable data. Several rigid registrationtechniques are implemented in commercial solutions forhospitals. These techniques cannot take into accountany non-rigid deformations, which may occur e. g. because of respiratory motion, the beating of theheart or shifts of the patient. This book shows theuse of integrating anon-rigid registration approach into the process ofreconstructing SPECT datasets of the myocardium. Theresults of the optimization of this process, as well as itsapplication to clinical images are discussed. Besides an overview of related works and patents inthe field of non-rigid image registration, this bookpresents the theoretic background needed tounderstand the concepts and methods behind this kindof registration.