Towards Quantifying Neurovascular Resilience

TL;DR

This paper presents a graph-based simulation method to quantify neurovascular resilience, enabling rapid, scalable analysis of vascular network robustness and potential biomarkers for adverse event risk prediction.

Contribution

It introduces a novel, efficient graph-based approach with analog circuits for simulating patient-specific neurovascular networks and assessing their resilience to pathological perturbations.

Findings

Effective simulation of pressure and flow in healthy and abnormal networks

Identification of biomarkers indicating network resilience

A graph sampling strategy for topological inference

Abstract

Whilst grading neurovascular abnormalities is critical for prompt surgical repair, no statistical markers are currently available for predicting the risk of adverse events, such as stroke, and the overall resilience of a network to vascular complications. The lack of compact, fast, and scalable simulations with network perturbations impedes the analysis of the vascular resilience to life-threatening conditions, surgical interventions and long-term follow-up. We introduce a graph-based approach for efficient simulations, which statistically estimates biomarkers from a series of perturbations on the patient-specific vascular network. Analog-equivalent circuits are derived from clinical angiographies. Vascular graphs embed mechanical attributes modelling the impedance of a tubular structure with stenosis, tortuosity and complete occlusions. We evaluate pressure and flow distributions, simulating healthy topologies and abnormal variants with perturbations in key pathological scenarios. These describe the intrinsic network resilience to pathology, and delineate the underlying cerebrovascular autoregulation mechanisms. Lastly, a putative graph sampling strategy is devised on the same formulation, to support the topological inference of uncertain neurovascular graphs.