Personalized Evaluation of the Passive Myocardium in Ischemic Cardiomyopathy Via Computational Modeling Using Bayesian Optimization Publisher Pubmed



Torbati S^{1, 2} ; Daneshmehr A² ; Pouraliakbar H³ ; Asgharian M⁴ ; Ahmadi Tafti SH^{1, 5} ; Shumtim D⁶ ; Heidari A^{4, 7, 8} Show Affiliations
Source: Biomechanics and Modeling in Mechanobiology Published:2024

Abstract

Biomechanics-based patient-specific modeling is a promising approach that has proved invaluable for its clinical potential to assess the adversities caused by ischemic heart disease (IHD). In the present study, we propose a framework to find the passive material properties of the myocardium and the unloaded shape of cardiac ventricles simultaneously in patients diagnosed with ischemic cardiomyopathy (ICM). This was achieved by minimizing the difference between the simulated and the target end-diastolic pressure–volume relationships (EDPVRs) using black-box Bayesian optimization, based on the finite element analysis (FEA). End-diastolic (ED) biventricular geometry and the location of the ischemia were determined from cardiac magnetic resonance (CMR) imaging. We employed our pipeline to model the cardiac ventricles of three patients aged between 57 and 66 years, with and without the inclusion of valves. An excellent agreement between the simulated and the target EDPVRs has been reached. Our results revealed that the incorporation of valvular springs typically leads to lower hyperelastic parameters for both healthy and ischemic myocardium, as well as a higher fiber Green strain in the viable regions compared to models without valvular stiffness. Furthermore, the addition of valve-related effects did not result in significant changes in myofiber stress after optimization. We concluded that more accurate results could be obtained when cardiac valves were considered in modeling ventricles. The present novel and practical methodology paves the way for developing digital twins of ischemic cardiac ventricles, providing a non-invasive assessment for designing optimal personalized therapies in precision medicine. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.