In recent years the interest in computational fluid dynamics (CFD)-based analysis of blood flow in coronary arteries has intensified, largely motivated by the possibility of non-invasive estimation of fractional flow reserve (FFR). FFR is important in the management of heart disease as it indicates whether there is insufficient blood flow entering the myocardium that would indicate a surgical intervention to restore sufficient flow. There have been various CFD-based modelling methodologies proposed in the literature with all of them crucially depending on the fidelity of coronary vasculature reconstruction from medical images - a process known as image segmentation. In this contribution, we describe an evaluation methodology of coronary artery segmentations developed by Hemolens Diagnostics and the objective assessment of its accuracy based on B-Spline transform. The results indicate that this methodology is an effective way to evaluate the model’s robustness to distortions.