摘要
阿尔茨海默症(AD)是一种在老年人中多发的脑部神经疾病,致病原因迄今未明,在疾病发展早期难以诊断。随着计算机和人工智能技术的大力发展,利用磁共振成像(MRI)技术和机器学习方法辅助医生对AD进行辅助诊断不断取得新的成果。本研究提出一种基于支持向量机递归特征消除(SVM-RFE)和线性判别分析(LDA)的AD辅助诊断方法。首先对MRI图像进行预处理,获得90个大脑脑区的灰质体积;然后使用SVM-RFE和LDA相结合的方法,对90个大脑脑区灰质体积进行特征选择;最后通过SVM进行分类。通过对来自于ADNI数据库中的34名AD、26名主观记忆衰退(SMC)患者和50名正常被试(NC)的MRI图像分析,得到AD/NC、AD/SMC和NC/SMC的平均分类准确率分别为94.0%、100.0%和93.6%。实验结果证明,本研究提出的方法可有效提取样本特征,辅助医生诊断AD。
Alzheimer’s disease(AD)is a brain neuropathy which is common among the aged.The pathogenesis of AD has not been known so far,and it is difficult to diagnose in the early stage of disease development.With the vigorous development of computers and artificial technologies,using magnetic resonance imaging(MRI)and machine learning methods to assist the diagnosis of AD has continuously made some new achievements.Herein a new method based on support vector machine-recursive feature elimination(SVM-RFE)and linear discriminant analysis(LDA)for the auxiliary diagnosis of AD is proposed.Firstly,the MRI images are preprocessed to obtain the gray matter volumes of 90 brain regions.Then the method combining SVM-RFE and LDA is used to select the significant features of the above gray matter volumes,and finally,the selected features are classified by SVM.By analyzing the MRI images of 34 patients with AD,26 patients with subjective memory complaints(SMC)and 50 normal controls(NC)from ADNI database,the average classification accuracies of AD/NC,AD/SMC and NC/SMC reach 94.0%,100.0%and93.6%,respectively.The experimental results show that the proposed method can effectively extract features and assist doctors in the diagnosis of AD.
作者
刘茜
王瑜
付常洋
肖洪兵
邢素霞
LIU Xi;WANG Yu;FU Changyang;XIAO Hongbing;XING Suxia(School of Computer and Information Engineering,Beijing Technology and Business University,Beijing 100048,China)
出处
《中国医学物理学杂志》
CSCD
2020年第5期656-660,共5页
Chinese Journal of Medical Physics
基金
国家自然科学基金(61671028)
2020年研究生科研能力提升计划项目。
关键词
阿尔茨海默症
磁共振成像
支持向量机
递归特征消除
线性判别分析
Alzheimer’s disease
magnetic resonance imaging
support vector machine
recursive feature elimination
linear discriminant analysis