Spatio-temporal registration and modelling of

Cardiovascular diseases are the single most important cause of death in the developed world. Their early diagnosis and treatment is becoming crucial in order to reduce mortal-ity and to improve the patient’s quality of life. The detection of cardiac diseases requires advanced methods for the quantification and analysis of the cardiovascular system. Re-cent advances in image modalities allow the high resolution imaging of the heart. Among them, MR imaging plays an increasingly important role for the understanding of the heart and the detection of cardiac abnormalities. However, MR imaging remains a new tech-nology limited to specialised imaging centers. Therefore, algorithms assisting with the interpretation of cardiac MR images are of high importance. In this thesis, several novel approaches for the spatio-temporal alignment of cardiac MR image sequences have been developed. The registration algorithms have the ability to correct any spatial misalignment caused by differences in the acquisition of the hearts and by local shape differences. Furthermore, the registration techniques have the ability to correct temporal misalignment caused by differences in the length of the cardiac cycle