Analysis of multimodal neuroimaging data for classification in mild cognitive impairment

摘要：

　　The accurate diagnosis of mild cognitive impairment(MCI)becomes more important,as the elderly is affected by Alzheimers disease(AD)widely.Many neuroimaging technologies have been used in pattern classification,e.g.structural and functional magnetic resonance imaging(MRI),positron emission tomography(PET).In this study,we classified MCI from healthy controls(HCs)using data from structural MRI(sMRI),18F-fluorodeoxyglucose(FDG)PET and 18F-florbetapir(AV45)PET,based on the multi-modal support vector machine(SVM)method.The data of three modalities were downloaded from the Alzheimers Disease Neuroimaging Initiative(ADNI)database including 130 subjects(72 MCI patients and 58 HCs).We extracted features from 90 regions of interest(ROIs)of gray matter(GM)and standardized uptake value ratios(SUVR)images,separately.Then,a classifier was built using 270 features of each subject based on the multi-modal SVM method.The accuracy,sensitivity and specificity of this classification were 70.1%,74.5%,and 67.6%,respectively.The results of the multi-modal SVM method are markedly superior to that of each single modality.The identification between MCI and HCs is helpful for the clinical diagnosis.

关键词： mild cognitive impairment sMRI PET classification support vector machine

作者: Ye Zhan Xia Wu Li Yao Kewei Chen Xiaojuan Guo

作者单位: College of Information Science and Technology,Beijing Normal University,Beijing,China,100875 College of Information Science and Technology,Beijing Normal University,Beijing,China,100875;State K Banner Alzheimers Institute and Good Samaritan PET Center,Phoenix,Arizona,USA,AZ 85006

会议类型: 国际会议

会议名称: The 2014 ICME International Conference on Complex Medical Engineering (CME2014)ICME复合医学工程国际会议

会议地点: 台北

会议语种:英文

页码: 3-7,41

在线出版日期: 2014-06-26（万方平台首次上网日期，不代表论文的发表时间）