Abstract:
The majority of patients with angina or heart failure have coronary artery disease. Bifurcations in the coronary arteries are particularly susceptible to pathological narrowing. Flow is a major factor of atheroma development, but limitations in imaging technology such as spatio-temporal resolution, signal-tonoise ratio (SNR), and imaging artefacts prevent in vivo investigations. Computational fluid dynamics (CFD) modelling is a common numerical approach, but it requires a sophisticated problem formulation. Left main bifurcation angles of 40°, 80° and 110° were selected, representing the spread of an atlas based population data of 100 computed tomography angiograms. Three bifurcations were reconstructed with 1) idealized, 2) stented, and 3) patient-specific geometry, and rapid prototyped to large scale phantoms. Their flow was reproduced using a blood-analogous, dynamically scaled steady flow circuit, enabling in vitro phase-contrast magnetic resonance (PC-MRI) measurements. Threshold segmentation and coherent point drift co-registration yielded a good result (σ2 <5.8x10-4) and after a natural-neighbour interpolation, the CFD and PC-MRI derived flow fields could be compared, showing very good agreement in magnitude (error 2-12%) and directional changes (r2 0.87 - 0.91) with the stented geometry correlating the best. PC-MRI over-estimated velocities close to the wall, possibly due to partial voluming. Bifurcation shape governed development of slow flow regions, which created lower SNR regions. These low SNR regions can likely be minimised in future by testing different similarity parameters to minimise acquisition error and improve correlation further. It was demonstrated that in vitro dynamic similarity acquisition correlates to true scale coronary flow simulations, and thus can overcome current PC-MRI’s spatio-temporal limitations. This novel method may elevate CFD coronary flow simulations in future by providing sophisticated boundary conditions using dynamically scaled in vitro PC-MRI.
