基于个体特异性功能连接的阿尔茨海默病早期识别研究  

Identification of Alzheimer’s disease and mild cognitive impairment patients using individual-specific functional connectivity

作　　者：王雪彤 董晓熹 李淑宇[1] WANG Xuetong;DONG Xiaoxi;LI Shuyu(School of Biological Science&Medical Engineering,Beijing Advanced Innovation Center for Biomedical Engineering,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学生物与医学工程学院,北京生物医学工程高精尖创新中心,北京100191

出　　处：《磁共振成像》2022年第4期56-61,68,共7页Chinese Journal of Magnetic Resonance Imaging

基　　金：国家自然科学基金(编号:81972160)。

摘　　要：目的基于静息态功能磁共振成像(resting state functional magnetic resonance imaging,rs-f MRI)探索个体特异性功能连接对阿尔茨海默病(Alzheimer’s disease,AD)及轻度认知障碍(mild cognitive impairment,MCI)患者、稳定型轻度认知障碍(stable mild cognitive impairment,sMCI)及进展型轻度认知障碍(progress mild cognitive impairment,pMCI)患者分类的影响,提取有助于AD及MCI诊断的潜在神经影像学标志物。材料与方法使用阿尔茨海默病神经影像学计划(Alzheimer’s Disease Neuroimaging Initiative,ADNI)数据集,包含47名正常对照组(normal controls,NC),66名s MCI,24名p MCI和29名AD患者。本文使用基于多任务学习的稀疏凸松弛交互结构优化(multi-task learning-based sparse convex alternating structure optimization,MTL-s CASO)方法提取个体特异性功能连接,并通过最小绝对收缩和选择算子(least absolute shrinkage and selection operator,LASSO)进行特征选择,最后利用支持向量机(support vector machine,SVM)分类器完成AD/MCI/NC的三分类及s MCI/p MCI的二分类任务。此外,采用双样本t检验来计算分类过程中最具辨识力的功能连接的组间差异(P<0.05)。结果相比于通过传统皮尔森相关构建的功能连接的分类结果(73.49%),基于个体特异性功能连接对AD/MCI/NC的三分类准确度达到了85.54%。此外,使用个体特异性功能连接对sMCI/pMC的分类性能(86.67%)要优于使用皮尔森相关得到的功能连接的分类性能(75.56%)。在分类过程中最具辨识力的功能连接,其连接强度在组间的差异有统计学意义。结论采用蕴含更多个体特性的个体特异性连接可提高对AD及MCI识别准确度,个体特异性功能连接有望作为AD及MCI诊断的潜在神经影像学标志物。Objective:To explore the value of the individual-specific functional connectivity based on resting state functional magnetic resonance imaging(rs-fMRI)in the classification of Alzheimer’s disease(AD)and mild cognitive impairment(MCI),and stable mild cognitive impairment(sMCI)and progress mild cognitive impairment(pMCI)patients.Materials and Methods:We used ADNI dataset,which included 47 normal controls(NC),66 sMCI,24 pMCI,and 29 AD patients.The individual-specific functional connectivity was used as input to select features with least absolute shrinkage and selection operator(LASSO).And SVM was performed for AD/MCI/NC multiclassification and s MCI/pMCI classification.We extracted the most discriminative functional connectivities,the two-sample t-test(P<0.05)was used to compare the differences in the strength of the most discriminative functional connectivities between groups.Results:Compared with the functional connectivity estimated by Pearson correlation(73.49%),the individual-specific functional connectivity estimated by multi-task learning-based sparse convex alternating structure optimization(MTL-sCASO)achieved 85.54%accuracy rate for AD/MCI/NC multiclassification.The individual-specific functional connectivity showed higher accuracy for identification sMCI and pMCI than the functional connectivity constructed by Pearson correlation(86.67%vs.75.56%).The strength of the most discriminative connectivity were significantly different between groups.Conclusions:The individual-specific connectivities is beneficial to the classification of AD and MCI,and the strength of functional connectivity could be used as a neuroimaging biomarker for the diagnosis of AD and MCI.

关 键 词：静息态功能磁共振成像 阿尔茨海默病 轻度认知障碍 多任务学习 个体特异性功能连接 早期诊断 影像学标志物 

分 类 号：R445.2[医药卫生—影像医学与核医学] R749.16[医药卫生—诊断学]