Content area
Abstract
Recent developments and sophistication of medical imaging technologies, such as Magnetic Resonance (MR), Computed Tomography and Ultrasonography, resulted in more accurate and detailed images of the human body, from which a wide range of valuable anatomical and physiological information can be extracted. Cardiac research has increasingly recognized that heart modeling can help interpret and analyze the provided information from imaging technologies. By combining it with prior biophysical information and other clinical data it is possible to obtain a more realistic model, which can be used for medical diagnosis, to predict surgical outcomes and finally to support treatment decisions. The heart is an organ with a complex structure and function, presenting involuntary cyclical movements, which together with the respiratory motion complicates image acquisition and cardiac modeling. Nevertheless, considering the extreme negative impact that cardiovascular diseases have in the society, the interest in cardiac modeling has received increasing attention, especially in the context of patient-specific medical simulations.
The present dissertation establishes a simple methodology to reconstruct the spatio-temporal left ventricle (LV) anatomy from MRI data for personalized cardiac simulations. Considering the main role that the LV plays in the cardiovascular system, the characterization of its anatomy and function is essential in the clinical practice for the diagnosis of pathological situations, such as the aortic valve diseases. Therefore, the development of personalized in-silico models that allow not only the observation of this heart chamber in 3D and its deformation along the cardiac cycle, but also enables the conduction of patient-specific cardiac simulations, can have an enormous impact on the diagnosis and treatment of cardiovascular diseases. The methodology here proposed extracts personalized geometric surfaces that represent accurately the anatomy of the inner LV surface in different phases of the cardiac cycle from MR images. In order to overcome the main challenges inherent to left ventricle modelling, such as the reconstruction of the apical and basal regions and the correction of motion-related artifacts, interactive techniques that combine different types of MR images used in clinical practice were developed. As final result, a set of smooth surfaces with high quality finite element meshes that represent accurately the geometric shape of the LV cavity in different phases of the cardiac cycle were obtained. This set of patient-specific LV surfaces describe the natural motion of a beating heart and encode information about the regions of the surface that deform at most between two instants of the cardiac cycle. The reconstructed shape of the LV with a smooth geometry and isotropic triangular mesh can be combined with other cardiac models to conduct patient-specific cardiac simulations, such as blood-flow simulations in patients with an aortic valve disease.
Concepts regarding the cardiac anatomy, function and modeling that were crucial for the understanding and development of this project will be addressed. This dissertation also presents the state-of-the-art in the field of LV reconstruction for simulations purposes, by introducing and discussing the most important works on this subject, giving special attention to recent projects based on MRI.





