Multi-resolution modeling of the mitral valve : a novel computational pipeline for patient-specific simulations of valve repair

The mitral valve (MV) is the left atrio-ventricular heart valve that regulates blood flow direction during the cardiac cycle. Among the four heart valves, MV is the most problematic one, with MV-related pathologies directly afflicting 5% of the population in the industrialized world. Over the past 25 years, computational simulations of the MV based on biomechanical models have gained significant credibility in understanding valve function and improving surgical treatments. However, MV models with proven predictive power have yet to be developed on a patient-specific basis from clinical imaging data. The main challenge is that ultrasound, which is the prevailing imaging modality in the clinic, struggles to capture the full MV shape and its fine-scale geometric details. Thus, computational modeling of the MV for clinical applications first requires overcoming the obstacle that complete MV models cannot be developed directly from clinical images. In this Ph.D. project, we tackled this challenge through a detailed anatomical analysis of the MV constituents to better understand the comprising components of the MV apparatus and their impact on the MV modeling. This knowledge was then used to systematically identify the key characteristic of predictive MV modeling, build patient-specific models, and perform simulations of the MV repair. Remarkably, we established a framework to build faithful computational models of the MV for predictive surgical simulations based only on the information that can be acquired in the clinic and prior to the operation.