数据驱动下合并狭窄左冠状动脉瘤血流动力学参数反演方法  

Data-Driven Inversion of Hemodynamic Parameters for Combined Stenotic Left Coronary Artery Aneurysms

作　　者：石政加 孙丽芳[2] 赵明轩 纪猛强 史玉龙 桑建兵[1] SHI Zhengjia;SUN Lifang;ZHAO Mingxuan;JI Mengqiang;SHI Yulong;SANG Jianbing(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300401,China;Department of Internal Medicine,Hebei University of Technology Hospital,Tianjin 300401,China)

机构地区：[1]河北工业大学机械工程学院,天津300401 [2]河北工业大学医院内科,天津300401

出　　处：《医用生物力学》2024年第5期853-859,共7页Journal of Medical Biomechanics

基　　金：河北省自然科学基金项目(A2020202015,A2021202014);国家自然科学基金项目(12102123)。

摘　　要：目的探究应用机器学习预测合并狭窄左冠状动脉瘤血流动力学参数预测。方法首先根据临床统计的合并狭窄左冠状动脉瘤几何参数范围进行参数化建模和仿真,将得到的仿真数据作为数据集,通过搭建两种常见的机器学习模型并训练优化,对壁面剪切力(wall shear stress,WSS)和压力这两个关键的血流动力学参数进行预测反演。通过对比分析这些模型在训练集和测试集上的表现,评估各个模型的准确性,验证数据驱动下合并狭窄左冠状动脉瘤血流动力学参数预测的有效性。结果确定了机器学习方法在动脉瘤血流动力学参数反演的有效性。对于WSS预测,训练后深度学习模型和随机森林模型的均方误差(mean squared error,MSE)、平均绝对误差(mean absolute error,MAE)、决定系数R^(2)分别达到了0.0528、0.0322、0.9883和0.0782、0.0463、0.9766。对于压力预测,深度学习模型和随机森林模型预测精度相当,MSE、MAE、R^(2)分别为4.67×10^(-6)、3×10^(-4)、0.9997和1.07×10^(-5)、5×10^(-4)、0.9993。结论机器学习方法在预测合并狭窄冠状动脉瘤模型的血流动力学参数方面表现出较高的精度,在进行机器学习预测时需要综合考虑模型的预测准确性、计算效率以及应用场景的需求,根据具体情况选择合适的模型。研究解果具有一定的临床意义,有助于医生更准确地评估患者病情,为心血管疾病的诊疗提供新思路和方法。Objective To investigate the application of machine learning to predict the hemodynamic parameters of combined stenotic left coronary artery(LCA)aneurysms.Methods Parameterized modeling and simulation based on the geometric parameter range of combined stenosis LCA aneurysms in clinical statistics were conducted.The obtained simulation data was used as the dataset,and two common machine learning models were constructed and trained for optimization to predict two key hemodynamic parameters:wall shear stress(WSS)and pressure.By comparing and analyzing the performances of these models on the training and testing sets,the accuracy of each model was evaluated,and the effectiveness of the data-driven prediction of hemodynamic parameters for LCA aneurysms with concomitant stenosis was verified.Results The effectiveness of machine learning method in inverting the hemodynamic parameters of aneurysms was determined.For WSS prediction,the trained deep learning model and random forest model achieved mean squared error(MSE),mean absolute error(MAE),and determination coefficient R^(2)of 0.0528,0.0322,0.9883,and 0.0782,0.0463,and 0.9766,respectively.For pressure prediction,the accuracies of the deep learning models and random forest models were comparable,with MSE,MAE,and R^(2)of 4.67×10^(-6),3×10^(-4),0.9997,and 1.07×10^(-5),5×10^(-4),and 0.9993,respectively.Conclusions Machine learning methods show high accuracy in predicting the hemodynamic parameters of combined stenotic coronary artery aneurysm models.The predictive accuracy of the model,computational efficiency,and needs of the application scenarios need to be considered in machine learning prediction so that the appropriate model can be selected according to the specific situation.This study has clinical significance,helping doctors to more accurately evaluate a patient’s condition and provide new ideas and method for the diagnosis and treatment of cardiovascular diseases.

关 键 词：动脉瘤 数值模拟 机器学习 深度学习 随机森林 血流动力学 

分 类 号：R318.01[医药卫生—生物医学工程]