Virtual Intervention of Intracranial Aneurysms

点击翻译按钮获取中文摘要

 DESCRIPTION (provided by applicant): Endovascular intervention is the predominant mode of for treating intracranial aneurysms (IAs). As a minimally invasive alternative to open-skull surgery, it obliterates an aneurysm by either filling it with platinum coils to decrease inflow and induce aneurysmal thrombosis, or diverting blood flow away using stent-like flow diverters (FDs) to induce gradual aneurysmal occlusion and parent vessel reconstruction. Despite its immense success, 30% of coiled IAs experience recanalization (recurrence), while 10% of FD-treated IAs fail to occlude. Patients experiencing such negative outcomes are subjected to increased risks for IA rupture and complications from treatment. This grant aims at developing a method to predict treatment outcome a priori. Our central hypothesis is that, with other factors, postprocedural hemodynamics predicts endovascular treatment outcome. This proposal aims to both develop clinically-practical computational tools to simulate endovascular treatment strategies and test the above hypothesis by creating predictive models that utilize hemodynamics from computational fluid dynamics (CFD) simulations on cases treated in silico. In Aim 1, we will develop and test rapid simulation tools for coil and FD implantation. Our methods are based on novel ball-winding (coil deployment) and ball-sweeping (FD deployment) algorithms. These methods improve upon existing ones by mimicking clinical deployment strategies with superior computational efficiency. To test if our modeling techniques recapitulate the effects of actual device deployment, we will compare CFD results from treated IAs in silico against hemodynamics experimentally measured by particle image velocimetry in treated patient- specific IA phantoms. In Aim 2, we will test the hypothesis that postprocedural hemodynamics, with other clinical factors, predicts patient angiographic outcome. To this end we will apply virtual intervention retrospectively to 700 treated IA cases at our institute, model post-treatment hemodynamics using CFD, and develop multivariate statistical models for treatment outcome based on patient data. We will use an innovative two-tiered statistical approach to extract models for treatment outcome prediction: discriminant function analysis to pre-screen a large number of candidate variables, followed by multivariate logistic regression for creation of parsimonious predictive models. In Aim 3, we will independently test the models prospectively on a new cohort of 300 treated IAs to determine if the models can correctly predict treatment outcome at 12 months. Successful completion of this project will establish-for the first time-a computational tool to predict IA treatment outcome a priori, thereby enabling neurosurgeons to assess different treatment strategies prior to device deployment. When implemented in the procedure room, this new ability will allow for optimization of treatment for individual patients and development of new strategies for those cases with higher failure rates. This project brings together experienced investigators from multiple disciplines and provides an unprecedented opportunity to translate engineering and computational advancements into clinical usage.