Personalized biventricular mechanics and sensitivity to model morphology

Abstract

AO_SCPLOWBSTRACTC_SCPLOWWe present a computational framework for constructing patient-specific models of cardiac mechanics based on standard clinical data, including electrocardiogram (ECG), cuff blood pressure, and electrocardiography-gated computed tomography angiography (CTA) imaging. The model is coupled to a closed-loop lumped parameter network (LPN) circulatory model and incorporates rule-based fiber architecture, as well as spatially varying epicardial boundary conditions to approximate surrounding tissue support. Model parameters are personalized through a multistep procedure that sequentially tunes circulatory dynamics, passive mechanics, and active contraction. The resulting personalized BiV model closely matches clinical pressure and volume measurements and reasonably agrees with image-based myocardial deformation. To assess the impact of anatomical model choice, we compare the BiV model to two commonly-used simplifications: a truncated BiV (t-BiV) model cut at the basal plane and a left ventricle-only (LV) model. For these models, we also evaluate their sensitivity to plausible variations in boundary conditions and contractile strength. With all other inputs held fixed, the LV model exhibits similar global pressure/volume behavior, despite moderate differences in regional deformation. In contrast, the t-BiV model produces substantial differences in both global function and local myocardial mechanics. These results suggest that while LV-only models may be sufficient for biomechanical studies, truncation at the basal plane strongly impacts model outputs and should be used with caution.