IMAGE-DRIVEN COMPUTATIONAL MODELS OF THE HEART FOR TETRALOGY OF FALLOT

In the last decades, researchers have been striving to develop computational models of the beating heart. Shifting from model generality to patient specificity, recent studies are demonstrating the potential impacts of such models on the clinical workflow. This chapter introduces three image-driven computational models of the heart that combine statistical and physiological priors for diagnosis, prognosis and therapy planning in repaired tetralogy of Fallot (rToF), a severe congenital heart disease. We first illustrate how physiological priors about the cardiac mechanics make the estimation of myocardium strain more reli- able, thus improving disease diagnosis. An algorithm that automatically tracks the heart along image sequences is constrained to estimate elastic and incompressible deformations, two fundamental properties of the myocardium. Then, we estimate a generative model of the right ventricular (RV) remodelling in rToF patients for disease prognosis. Computed us- ing statistical shape analyses and partial least squares, the model suggested that the dilation, the basal bulging and the apical dilation typically observed in these patients appear progres- sively as the child grows. These findings could support the cardiologist in predicting the evolution of the pathology for planning pulmonary valve replacement (PVR), the current state-of-the-art therapy in rToF. Finally, we introduce an electromechanical (EM) model of the heart for personalised planning of PVR with RV volume reduction in two patients. The EM model simulates the main features of the beating heart. After personalisation, the virtual heart is used to simulate PVR. As expected, the predicted postoperative function significantly improved in both patients. As illustrated by these results, combining medical