Related Experiment Video
Updated: Aug 24, 2026
Comprehensive Endovascular and Open Surgical Management of Cerebral Arteriovenous Malformations
Published on: October 20, 2017
Clinical outcome prediction of Contour Neurovascular System treatment based on pre-treatment geometric and
Fina Gießler1, Anna Bernovskis2,3, Franziska Gaidzik2,4
1Institute for Medical Informatics and Statistics, University Hospital Schleswig-Holstein, Kiel, Germany. giessler@medinfo.uni-kiel.de.
Purpose:
The contour neurovascular system (CNS) is increasingly used for endovascular treatment of intracranial aneurysms, but outcome and post-treatment device behavior vary between patients. This study investigated whether pre-treatment geometric and hemodynamic characteristics from image-based blood flow simulations are associated with clinical outcome and follow-up CNS shape modifications.
Methods:
Pre-treatment vascular geometries and blood flow simulations were analyzed in 39 CNS-treated cases. Geometric, hemodynamic, and ostium-flow clustering features were extracted, followed by global feature thinning to reduce redundancy. Associations with binary clinical outcome and follow-up shape modifications (SM1-SM3) were assessed using nonparametric univariate statistics with false discovery rate correction. Exploratory outcome discrimination was evaluated using a support vector machine classifier.
Results:
Aneurysm geometry showed the strongest association with clinical outcome. Smaller dome and ostium areas were associated with positive outcome (Cliff's δ = - 0.57 and - 0.43), while hemodynamic parameters showed weaker but directionally consistent trends, including lower peak outflow velocity, and reduced inflow intensity. Shape-modification patterns differed across deformation classes. SM1 and SM2 showed no FDR-significant features but moderate effects for aneurysm geometry and selected hemodynamic metrics. SM3 showed the strongest associations with oscillatory flow characteristics (δ ≈ 0.56-0.57) and peripheral outflow dominance. Exploratory machine learning showed limited, non-significant outcome discrimination (mean AUC = 0.57), with selected features consistent with univariate findings.
Conclusion:
Pre-treatment geometric and hemodynamic characteristics are linked not only to clinical outcome but also to post-treatment CNS shape modifications. By linking pre-operative flow conditions to follow-up device behavior, this study improves the understanding of hemodynamic factors associated with CNS shape modifications and suggests that simulation-based analysis may support future patient-specific treatment planning using exclusively pre-operative data.
