Automated Method for the Volumetric Evaluation of Myocardial Scar from Cardiac Magnetic Resonance Images

In most western countries cardiovascular diseases are the leading cause of death, and for the survivors of ischemic attack an accurate quantification of the extent of the damage is required to correctly assess its impact and for risk stratification, and to select the best treatment for the patient. Moreover, a fast and reliable tool for the assessment of the cardiac function and the measurement of clinical indexes is highly desirable. The aim of this thesis is to provide computational approaches to better detect and assess the presence of myocardial fibrosis in the heart, particularly but not only in the left ventricle, by performing a fusion of the information from different magnetic resonance imaging sequences. We also developed and provided a semiautomatic tool useful for the fast evaluation and quantification of clinical indexes derived from heart chambers volumes. The thesis is composed by five chapters. The first chapter introduces the most common cardiac diseases such as ischemic cardiomyopathy and describes in detail the cellular and structural remodelling phenomena stemming from heart failure. The second chapter regards the detection of the left ventricle through the development of a semi-automated approach for both endocardial and epicardial surfaces, and myocardial mask extraction. In the third chapter the workflow for scar assessment is presented, in which the previously described approach is used to obtain the 3D left ventricle patient-specific geometry; a registration algorithm is then used to superimpose the fibrosis information derived from the late gadolinium enhancement magnetic resonance imaging to obtain a patientspecific 3D map of fibrosis extension and location on the left ventricle myocardium. Focus of the fourth chapter is on the left atrium, and fibrotic tissue detection for gaining insight on atrial fibrillation. In the fifth chapter some conclusive remarks are presented with possible future developments of the presented work.