基于血流动力学与形态学特征融合的动脉瘤破裂风险机器学习预测研究

陈广新; 才莹; 郭金兴; 于成龙

doi:10.3969/j.issn.2096-6091.2023.23.002

您当前的位置：

首页 >

文章列表页 >

基于血流动力学与形态学特征融合的动脉瘤破裂风险机器学习预测研究

研究论文 | 更新时间：2024-10-08

- 基于血流动力学与形态学特征融合的动脉瘤破裂风险机器学习预测研究
- Machine Learning Prediction of Aneurysm Rupture Risk Based on the Fusion of Hemodynamic and Morphological Features
- 新一代信息技术 2023年6卷第23期页码：9-14
- 作者机构：
  
  1.牡丹江医学院医学影像学院，黑龙江牡丹江 157011
  2.牡丹江医学院附属红旗医院，黑龙江牡丹江 157011
- 作者简介：
  
  陈广新（1978—），男，黑龙江省哈尔滨市人，讲师，硕士研究生学历，研究方向：医学人工智能。
  才莹（1996—），女，黑龙江省东宁市人，主治医师，研究生学历，研究方向：医学人工智能。
  郭金兴（1984—），女，黑龙江省穆棱人，主管护师，研究方向：医学人工智能。
  于成龙
- 基金信息：
  
  黑龙江省卫委员会2022年科技（资助）项目(20220909040626);黑龙江省省属高校科研基本业务费科研项目(2023-KYYWF-0921)
- DOI：10.3969/j.issn.2096-6091.2023.23.002
  中图分类号： TP30
- 纸质出版日期：2023-12-15，
- 稿件说明：
移动端阅览
陈广新, 才莹, 郭金兴, 等. 基于血流动力学与形态学特征融合的动脉瘤破裂风险机器学习预测研究[J]. 新一代信息技术, 2023, 6(23): 09-14

CHEN Guang-xin, CAI Ying, GUO Jin-xing, et al. Machine Learning Prediction of Aneurysm Rupture Risk Based on the Fusion of Hemodynamic and Morphological Features[J]. New Generation of Information Technology, 2023, 6(23): 09-14
陈广新, 才莹, 郭金兴, 等. 基于血流动力学与形态学特征融合的动脉瘤破裂风险机器学习预测研究[J]. 新一代信息技术, 2023, 6(23): 09-14 DOI： 10.3969/j.issn.2096-6091.2023.23.002.

CHEN Guang-xin, CAI Ying, GUO Jin-xing, et al. Machine Learning Prediction of Aneurysm Rupture Risk Based on the Fusion of Hemodynamic and Morphological Features[J]. New Generation of Information Technology, 2023, 6(23): 09-14 DOI： 10.3969/j.issn.2096-6091.2023.23.002.

摘要

本研究旨在通过整合动脉瘤患者的血流动力学与形态学特征，构建一个预测动脉瘤破裂风险的机器学习模型。收集了2021年2月至2023年12月期间在牡丹江医学院附属红旗医院神经外科就诊的颅内动脉瘤患者的病例。研究组包括130例动脉瘤破裂患者，对照组包括60例未破裂的动脉瘤患者。所有患者均提供了DICOM格式的CT（Computed Tomography）图像，并在获得医院伦理委员会的批准和患者家属签署知情同意书后，使用MIMICS软件对动脉瘤的DICOM格式CT影像进行了三维重建。通过形态学测量，我们获得了形态学参数指标，并通过计算流体力学仿真计算得到了血流动力学参数。将形态学参数与血流动力学参数相结合，构建了一个用于机器学习临床预测模型的数据集。研究采用了10种机器学习算法构建预测模型，并使用准确率、AUC（Area Under Curve）、召回率和

分数等指标进行评估。此外基于SHAP方法进行模型可解释性分析。在所有测试的模型中，随机森林模型展现出最佳性能，其准确率达到了0.78，AUC值为0.81，且召回率也较高，达到了0.72。通过融合动脉瘤的形态学特征与血流动力学特征构建的机器学习模型，能够为临床决策提供有力的工具，并展现出良好的临床应用潜力。

Abstract

The aim of this study is to construct a machine learning model for predicting the risk of aneurysm rupture by integrating hemodynamic and morphological features of patients with intracranial aneurysms. Cases of patients with intracranial aneurysms who sought treatment at the Department of Neurosurgery

Affiliated Hongqi Hospital of Mudanjiang Medical University from February 2021 to December 2023 we

re collected. The study group comprised 130 patients with ruptured aneurysms

while the control group included 60 patients with unruptured aneurysms. All patients provided CT (Computed Tomography) images in DICOM format

and after obtaining approval from the hospital’s ethics committee and signed informed consent from the patients’ families

three-dimensional reconstruction of the aneurysm DICOM format CT images was performed using MIMICS software. Morphological parameters were obtained through morphological measurements

and hemodynamic parameters were calculated through computational fluid dynamics simulations. The dataset for constructing the machine learning clinical prediction model was built by combining morphological and hemodynamic parameters. Ten machine learning algorithms were employed to construct the prediction models

and their performance was evaluated using metrics such as accuracy

AUC (Area Under the Curve)

recall rate

and

score. Additionally

model interpretability was analyzed using the SHAP method. Among all the models tested

the random forest model demonstrated the best performance with an accuracy of 0.78

an AUC value of 0.81

and a high recall rate of 0.72. The machine learning model constructed by integrating the morphological and hemodynamic features of aneurysms can provide a powerful tool for clinical decision-making and shows promising potential for clinical application.

关键词

动脉瘤破裂风险血流动力学特征形态学特征机器学习模型可解释性

Keywords

aneurysmrupture riskhemodynamic featuresmorphological featuresmachine learningmodel interpretability

references

杨新健, 张鸿祺. 重视颅内未破裂动脉瘤的临床和基础研究[J]. 中华神经外科杂志, 2017, 33(8): 757-759.

LI M H, CHEN S W, LI Y D, et al. Prevalence of unruptured cerebral aneurysms in Chinese adults aged 35 to 75 years: A cross-sectional study[J]. Annals of Internal Medicine, 2013, 159(8): 514-521.

THOMPSON B G, JrBROWN R D, AMIN-HANJANI S, et al. Guidelines for the management of patients with unruptured intracranial aneurysms: A guideline for healthcare professionals from the American heart association/american stroke association[J]. Stroke, 2015, 46(8): 2368-2400.

WILLIAMS L N, JrBROWN R D. Management of unruptured intracranial aneurysms[J]. Neurology: Clinical Practice, 2013, 3(2): 99-108.

蔡明辰, 吕仁杰, 韩松林, 等. 颅内动脉瘤形成及破裂因素的研究进展[J]. 世界最新医学信息文摘, 2019, 19(61): 118.

DHAR S, TREMMEL M, MOCCO J, et al. Morphology parameters for intracranial aneurysm rupture risk assessment[J]. Neurosurgery, 2008, 63(2): 185-197.

MENG H, TUTINO V M, XIANG J, et al. High WSS or low WSS? Complex interactions of hemodynamics with intracranial aneurysm initiation, growth, and rupture: Toward a unifying hypothesis[J]. American Journal of Neuroradiology, 2014, 35(7): 1254-1262.

SHIMOGONYA Y, ISHIKAWA T, IMAI Y, et al. Can temporal fluctuation in spatial wall shear stress gradient initiate a cerebral aneurysm? A proposed novel hemodynamic index, the gradient oscillatory number (GON)[J]. Journal of Biomechanics, 2009, 42(4): 550-554.

FRÖSEN J, PIIPPO A, PAETAU A, et al. Remodeling of saccular cerebral artery aneurysm wall is associated with rupture: Histological analysis of 24 unruptured and 42 ruptured cases[J]. Stroke, 2004, 35(10): 2287-2293.

FRÖSEN J, CEBRAL J, ROBERTSON A M, et al. Flow-induced, inflammation-mediated arterial wall remodeling in the formation and progression of intracranial aneurysms[J]. Neurosurgical Focus, 2019, 47(1): E21.

王珏, 赵玉武, 李明华, 等. 多发性颅内动脉瘤破裂的危险因素分析[J]. 介入放射学杂志, 2014, 23(1): 1-4.

INVESTIGATORS UIA. Unruptured intracranial aneurysms: Risk of rupture and risks of surgical intervention-International study of unruptured intracranial aneurysms investigators[J]. The New England Journal of Medicine, 1998, 339(24): 1725-1733.

BRINJIKJI W, ZHU Y Q, LANZINO G, et al. Risk factors for growth of intracranial aneurysms: A systematic review and meta-analysis[J]. American Journal of Neuroradiology, 2015, 37(4): 615-620.

VARBLE N, TUTINO V M, YU J, et al. Shared and distinct rupture discriminants of small and large intracranial aneurysms[J]. Stroke, 2018, 49(4): 856-864.

LINDGREN A E, KOIVISTO T, BJÖRKMAN J, et al. Irregular shape of intracranial aneurysm indicates rupture risk irrespective of size in a population-based cohort[J]. Stroke, 2016, 47(5): 1219-1226.

KASHIWAZAKI D, KURODA S. Size ratio can highly predict rupture risk in intracranial small (<5 mm) aneurysms[J]. Stroke, 2013, 44(8): 2169-2173.

HOH B L, SISTROM C L, FIRMENT C S, et al. Bottleneck factor and height-width ratio: Association with ruptured aneurysms in patients with multiple cerebral aneurysms[J]. Neurosurgery, 2007, 61(4): 716-723.

ZHANG Y, JING L, LIU J, et al. Clinical, morphological, and hemodynamic independent characteristic factors for rupture of posterior communicating artery aneurysms[J]. Journal of Neurointerventional Surgery, 2016, 8(8): 808-812.

LIU J, FAN J X, XIANG J P, et al. Hemodynamic characteristics of large unruptured internal carotid artery aneurysms prior to rupture: A case control study[J]. Journal of Neurointerventional Surgery, 2016, 8(4): 367-372.

周涛, 刘赟璨, 侯森宝, 等. M~3 Res-Transformer: 新冠肺炎胸部X-ray图像识别模型[J]. 电子学报, 2024, 52(2): 589-601.

徐树生, 林孝工, 李新飞. 强跟踪自适应平方根容积卡尔曼滤波算法[J]. 电子学报, 2014, 42(12): 2394-2400.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于可解释性机器学习的阿尔兹海默症痴呆风险预测

基于人工智能的无人机自组网路由算法研究进展

预测COVID-19感染风险的医疗决策机器学习方法研究