摘要
目的为降低梯形截面丝支架植入颅内囊状动脉瘤后瘤腔破裂的风险,对支架丝截面底边长度进行优化设计。方法构建38种不同底边长度的梯形截面支架及其植入动脉瘤后的有限元模型,利用流固耦合数值模拟计算得到38组最大瘤腔壁面压力梯度值,并应用广义回归神经网络及遗传算法对梯形截面底边长度进行优化,使植入支架后最大瘤腔壁面压力梯度降到最低。结果优化结果显示,相对传统矩形截面支架,优化后支架将最大瘤腔壁面压力梯度降低了7.86%。结论广义回归神经网络与遗传算法结合可以很好地解决支架优化方面的问题。
Objective To optimize the baseline on the trapezoidal cross-section of stent wires, so as to reduce the risk of intracranial saccular aneurysm rupture after the implantation of such stents, Methods Thirty-eight trape- zoidal cross-section wire stents with different baselines were constructed to establish the finite element models. Numerical simulation by fluid-solid interaction method was conducted to calculate 38 maximum pressure gradients on the aneurysm wall. GRNN (general regression neural network) and GA (genetic algorithm) were used to op- timize the baseline on the cross-section of stents with trapezoidal cross-section wire so as to minimize the maxi- mal pressure gradient on the aneurysm wall. Results Compared with the traditional stent with rectangular crosssection wire, the maximal pressure gradient on the a neurysm wall was reduced by ?. 86% after the implantation with the optimized stent with trapezoidal cross-section wire. Conclusions The combination of GRNN and GA is an effective approach for stent optimization.
出处
《医用生物力学》
EI
CAS
CSCD
北大核心
2012年第3期294-298,共5页
Journal of Medical Biomechanics
基金
国家自然科学基金资助项目(10972016
81171107)
北京市自然科学基金资助项目(3092004)
关键词
支架
广义回归神经网络
遗传算法
压力梯度
有限元分析
数值模拟
Stent
General regression neural network (GRNA)
Genetic algorithm (GA)
Pressure gradient
Finite element analysis
Numerical simulation