Does the effort in replicating the 3D morphology of coronary arteries when deriving patient-specific computational models from intravascular imaging make sense?

Cardiovascular diseases are the leading cause of death in western countries. In coronary atherosclerosis, local hemodynamics and wall mechanics are key determinants of plaque evolution. Recently, combining computational tools with medical imaging has enabled to profile biomechanical quantities, such as wall shear stress (WSS) and wall stress, with high accuracy. Specifically, patient-specific computational fluid dynamics (CFD) simulations elucidated the role of WSS in coronary atherosclerosis natural history, while finite element analysis (FEA) quantified volumetric wall and plaque stress/strain under patho-physiological loading conditions. However, their clinical applicability is hindered by the assumptions related to the model-based strategy and by the lack of reliable automatic reconstruction methods from clinical imaging. Even with an accurate segmentation of intravascular images, the conversion of labeled frames into digital twin computational models presents several difficulties, such as operator-dependence affecting frame-to-frame registration, rotation and alignment to reproduce the in vivo orientations. In this context, the aim of this work was to assess the impact of in vivo coronary arteries 3D morphology replication on structural and hemodynamics computational models, with the final goal to reveal if the efforts in replicating the coronary in vivo morphology when deriving patient-specific computational models from intravascular imaging make sense. To do that, the coronary arteries (3 left anterior descending, and 3 right coronary arteries) of 6 patients were imaged by Optical Coherence Tomography (OCT) and Coronary Computed Tomography Angiography (CCTA). The patient-specific geometries were reconstructed aligning OCT-based wall component segmentations along the CCTA-based centerline (Real models). Six additional models were obtained aligning OCT-based segmentations along a straight centerline (Straight models). By construction, this approach could be fully automatized. The differences between the two modeling approaches were assessed through CFD and FEA simulations. Briefly, CFD simulations were performed by prescribing patient-specific inflow conditions, while FEA were conducted by applying luminal patient-specific systolic pressure, using an iterative backward-incremental method. Hemodynamic analysis was focused on near-wall WSS-based quantities, i.e., time-averaged wall shear stress (TAWSS), oscillatory shear index, relative residence time, and topologic shear variation index, and intravascular quantities, i.e., local normalized helicity and h-index. Structural analysis included wall structural stress investigation, through the Von Mises Stress (VMS). Across the six cases, Straight models emerged as not representative for WSS multidirectionality, markedly attenuated compared to Real geometries. Instead, WSS magnitude was preserved: comparable TAWSS luminal maps resulted for Straight vs. Real models. Intravascular descriptors reflected the absence of 3D curvature: Straight models exhibited lower helicity intensity and a more balanced distribution of helical structures than Real ones. As for FEA, although the resulting VMS presented a similar distribution patterns between Straight and Real models, the Straight one displayed marginally increased stress levels compared to the Real model, particularly close to the lipid zone. In conclusion, an accurate reproduction of full 3D coronary geometry is essential to properly capture curvature-dependent biomechanical effects.