Increasing Accessibility of Biomechanical Stress Analysis of Abdominal Aortic Aneurysms through Model Simplification
Eric Shang, MD1, Benjamin Jackson, MD2.
1West Virginia University, Morgantown, WV, USA, 2University of Pennsylvania, Philadelphia, PA, USA.
Objective: Previous studies have shown that peak wall stress (PWS) derived from finite element analysis (FEA) of abdominal aortic aneurysms (AAAs) predicts aneurysm growth and rupture better than diameter alone. One significant barrier to adoption of stress analysis is computational time and model complexity. However, there has not been evidence tying increasing model complexity to improved clinical prediction. We hypothesize that model simplification can reduce FEA computation time without significantly degrading its ability to predict clinical outcomes.
Methods: Patients with AAAs (n=35) undergoing radiologic surveillance were identified. Custom MATLAB algorithms generated aortic geometries from CTA images. A full model consisting of a variable wall thickness aortic wall, intraluminal thrombus, and wall calcifications each with their own specific material properties and interactions was generated. A simplified model was created consisting of fewer elements and ignoring the separate properties of thrombus and calcifications. These models were loaded with systolic blood pressure using FEA. PWS and aneurysm growth (as a proxy for rupture risk and the need for repair) were examined for both models.
Results: The average radiologic follow-up time was 23.6±12.9 months and the average aneurysm growth rate was 2.9±1.6 mm/year. Initial aortic diameter was not found to be significantly correlated with aneurysm growth (r=0.30, P=0.12). The simplified model had 76% fewer elements (1.7x106 vs 4.1x105), and had significantly decreased computation time (25.2±11.9 min vs 312±184min, P<0.001) on a commercially available 8-core workstation. PWS calculated by the both the full model (r=0.87, p<0.001) and the simplified model (r=0.79, p<0.001) were significantly correlated with aneurysm growth. A stronger correlation to
aneurysm growth was found in PWS generated by the simplified model as compared to aneurysm diameter (r=0.79 vs r=0.30, P=0.012 by Fisher’s r to Z transformation.)
Conclusions: Model simplification significantly decreased the computation time but preserved the ability of FEA calculated wall stress to predict AAA growth. Further efforts at improving accessibility of biomechanical risk stratification may lead to its adoption into clinical decision making.
Back to 2018 ePosters
