Worcester Polytechnic Institute
Mathematical Sciences Department
PhD Dissertation Defense
In Vivo IVUS-Based 3D Fluid-Structure Interaction Models for Human Coronary Atherosclerotic Plaque Vulnerability Assessment and Progression Prediction
Introduction Atherosclerotic plaque progression and vulnerability are believed to be associated with plaque morphology and mechanical conditions. Computational Fluid-Structure Interaction (FSI) models for human coronary plaques based on in vivo intravascular ultrasound (IVUS) data were constructed to obtain stress/strain and flow shear stress conditions for plaque progression and rupture investigations. New computational models with residual stress included were constructed to improve model accuracy. We are testing the hypothesis that combining plaque morphology and mechanical conditions could better predict plaque progression and vulnerability.
Data, Model and Method. Patient-specific one time-point and follow-up IVUS data of coronary plaques and X-ray angiography were provided by Dr. Mintz’s group at Cardiovascular Research Foundation, Dr. Zheng’s group at Washington University in St. Louis and Dr. Zhu’s group at Zhongda Hospital, Southeast University, Nanjing, China. Blood flow was assumed to be laminar, viscous, incompressible and Newtonian. The incompressible Navier-Stokes equations with arbitrary Lagrangian-Eulerian (ALE) formulation were used as the governing equations. The coronary vessel material was assumed to be hyperelastic, anisotropic, nearly-incompressible and homogeneous. Plaque components were assumed as isotropic. The Mooney-Rivlin models were used as the structure models. 3D FSI models were constructed to calculate stress/strain and flow conditions for our investigations. A three-step procedure consisting of wrapping, axial stretching and pressurizing was used to include residual stress in the model and improve the accuracy of stress/strain calculations. Computable plaque vulnerability indices (MPVI and SPVI) were introduced based on critical mechanical conditions and plaque morphological features to serve as alternatives to assess plaque vulnerability while true plaque vulnerability is not available. All possible combinations of plaque morphological characteristics (wall thickness, lipid percent, minimal cap thickness) and mechanical conditions (stress/strain and flow shear stress) were tested using the generalized linear mixed-effect models (GLMMs) to identify the best risk predictor(s) measured by their sensitivity and specificity predicting plaque vulnerability and progression.
Results. A combination of wall thickness (WT) and plaque wall stress (PWS) was identified as the best predictor for plaque progression measured by wall thickness increase (WTI) with sum of sensitivity and specificity acting as prediction accuracy. Plaque area increase (PAI) and plaque burden increase (PBI) were also used as other measurements to size plaque progression, and their best predictors are WT + PWS + PA + PB and WSS + PA, respectively.
Given the strong correlation (r=0.6168) and agreement rate (55.14%) between MPVI and SPVI, SPVI could be used as another index for assessing plaque vulnerability, complementing MPVI. Vulnerability prediction analysis was performed following plaque progression study to identify the best predictors for future plaque vulnerability measured by changes of MPVI and SPVI (MPVI and SPVI), respectively. Among all possible combinations of risks factors, the combination of minimum cap thickness, MPVI, plaque area, and SPVI was the best predictor for MPVI while the combination of minimum cap thickness, critical wall shear stress, plaque area and SPVI was the winner for SPVI.
Our residual stress model comparison study indicated that without residual stress, the lumen and cap stress could be overestimated by 400% or higher. However, stress/strain calculations from our FSI models can still be used since relative differences among patients are still meaningful.
Discussion and Conclusion. With large-scale patient study validations, our computable plaque vulnerability indices and identified potential risk factor(s) for plaque progression and future vulnerability will serve as valid guidance for proper clinical intervention.
Prof. Dalin Tang, Advisor – Department of Mathematical Sciences, WPI
Prof. Allen Hoffman – Department of Mechanical Engineering, WPI
Prof. Mayer Humi - Department of Mathematical Sciences, WPI
Prof. Roger Lui - Department of Mathematical Sciences, WPI
Prof. Zheyang Wu - Department of Mathematical Sciences, WPI