In-Silico Investigation of 3D Quantitative Angiography for Internal Carotid Aneurysms Using Biplane Imaging and 3D Vascular Geometry Constraints

Abstract

Quantitative angiography (QA) in two dimensions has been instrumental in assessing neurovascular contrast flow patterns, aiding disease severity and treatment outcome evaluations. However, QA requires high spatio-temporal resolution, restricting its use to digital subtraction angiography (DSA), and is prone to errors in quantification of highly 3D flow patterns. This study examines whether 3D QA information can be recovered by reconstructing four-dimensional (4D) angiography using data from standard clinical imaging protocols. Patient-specific internal carotid aneurysm models were used to generate high-fidelity computational fluid dynamics (CFD) simulations of contrast flow. The resulting 4D angiograms were used to simulate biplane DSA under clinical imaging protocols. 4D angiography was reconstructed from two views using back-projection constrained by an a priori 3D geometry. Quantitative angiographic parametric imaging (API) metrics obtained from the CFD-based 4D angiography and reconstructed 4D angiography were compared using mean square error (MSE) and mean absolute percentage error (MAPE). The reconstructed 4D datasets effectively captured 3D flow dynamics, achieving an average MSE of 0.007 across models and flow conditions. API metrics such as PH and AUC closely matched the CFD ground truth, with temporal metrics showing some variability in regions with overlapping projections. These results demonstrate the potential to recover 3D QA information using simulated 4D angiography constrained by standard clinical imaging parameters. The method provides a robust framework for evaluating and improving QA in clinical neurovascular applications, offering new insights into the dynamics of aneurysmal contrast flow.