神经突方向离散度和密度成像结合机器学习对遗忘型轻度认知障碍脑微观结构的研究

Studying on Brain Microstructure of Amnestic Mild Cognitive Impairment by Neurite Orientation Dispersion and Density Imaging Combined with Machine Learning

目的采用神经突方向离散度和密度成像(NODDI)结合机器学习算法,研究遗忘型轻度认知障碍(aMCI)脑微观结构的改变。方法搜集26例aMCI患者和24例健康对照者,进行磁共振扩散序列扫描,采用NODDI_toolbox对扩散数据进行处理,获得NODDI参数神经突密度指数(NDI)、方向离散度指数(ODI)和各向同性水分子体积分数(Viso),基于约翰霍普金斯大学脑白质图谱,制作11个脑白质结构的模板,包括双侧扣带束、胼胝体膝部和压部、双侧内囊后肢、双侧上纵束、双侧钩束和穹窿的模板,基于解剖学自动标记模板(AAL),制作14个脑灰质结构的模板,包括双侧海马、双侧海马旁回、双侧杏仁核、双侧尾状核、双侧苍白球、双侧壳核、双侧丘脑的模板,提取模板的NDI、ODI和Viso值,应用K最近邻(KNN)、Logistic回归(LR)、随机森林(RF)、支持向量机(SVM)机器学习算法评估NODDI各参数对aMCI的预测能力;并对提取值进行独立样本t检验,具有统计学差异的结构与简易智能状态检查量表(MMSE)和蒙特利尔认知评估量表(MoCA)评分进行相关性分析。结果采用KNN、LR、RF、SVM机器学习算法,所有模板NDI值预测aMCI的受试者工作特征(ROC)曲线下面积(AUC)分别是0.781、0.719、0.833、0.766,所有模板ODI值预测aMCI的AUC分别是0.891、0.859、0.914、0.922;所有模板Viso值预测aMCI的AUC分别是0.914、0.891、0.922、0.891;所有模板NODDI所有参数(NODDI_all)预测aMCI的AUC分别是0.938、0.922、0.969、0.953;在白质结构中,aMCI组有45.45%(5/11)的NDI值显著降低,18.18%(2/11)的ODI值显著降低,27.27%(3/11)的Viso值显著增加;在灰质结构中,aMCI组有7.14%(1/14)的NDI值显著降低,71.43%(10/14)的ODI值显著降低,14.29%(2/14)的Viso值显著增加;相关性分析,右侧钩束NDI值,左侧上纵束、右侧海马、左侧尾状核、左侧苍白球ODI值,双侧扣带束、左侧海马Viso值均与MMSE评分相关;胼胝体压部、双侧钩束、左侧杏仁核NDI值,左侧上纵束、双侧海马、双侧海马旁回、左侧尾状核、左侧苍白球、左侧壳核、双侧丘脑ODI值,双侧扣带束、左侧海马Viso值均与MoCA评分相关。结论MCI阶段脑白质微观结构以神经突密度降低为主,灰质微观结构以树突复杂性降低为主,NODDI可能能够反映aMCI的临床认知状态,NODDI结合机器学习算法,有望成为aMCI早期诊断的新方法。

Objective To investigate the changes of brain microstructure in amnestic mild cognitive impairment(aMCI)by using neurite orientation dispersion and density imaging(NODDI)combined with machine learning algorithms.Methods Twenty-six aMCI patients and 24 healthy controls were recruited and scanned by diffusion MR sequence.The diffusion data were processed by NODDI_toolbox and the NODDI parameters including neurite density index(NDI),orientation dispersion index(ODI)and volume fraction of isotropic water molecules(Viso),were obtained.Based on the white matter atlas of Johns Hopkins University,11 templates of white matter(WM)structures were made,including the bilateral cingulum,genu and splenium of corpus callosum,the bilateral posterior limb of internal capsule,bilateral superior longitudinal fasciculus,bilateral uncinate fasciculus and fornix.Fourteen templates of gray matter(GM)structures were made based on anatomical automatic labeling template(AAL),which included the bilateral hippocampus,bilateral parahippocampal gyrus,bilateral amygdala,bilateral caudate nucleus,bilateral globus pallidus,bilateral putamen and bilateral thalamus.The NDI,ODI and Viso values of all the templates were extracted.The machine learning algorithms,such as K nearest neighbor(KNN),logistic regression(LR),random forest(RF)and support vector machine(SVM),were used to evaluate the diagnostic efficiency of each parameter values on aMCI.The extracted templates values were tested between the two groups by independent sample t test and the correlation between the structures with statistical difference and Mini Mental State examination(MMSE)and Montreal Cognitive Assessment(MoCA)scores were analyzed.Results Using the KNN,LR,RF and SVM machine learning algorithms,the area under the receiver operating characteristic(ROC)curve(AUC)of predicting aMCI by the NDI values of all the templates were 0.781,0.719,0.833 and 0.766 respectively;the AUCs of predicting aMCI by the ODI values of all the templates were 0.891,0.859,0.914 and 0.922 respectively;the AUC of predicting aMCI by the Viso values of all the templates were 0.914,0.891,0.922 and 0.891 respectively;the AUC of predicting aMCI by all the NODDI parameters of all the templates(NODDI_all)were 0.938,0.922,0.969 and 0.953 respectively.In the WM,the NDI values of 45.45%(5/11)and the ODI values of 18.18%(2/11)decreased,and the Viso values of 27.27%(3/11)increased in the aMCI group.While in the GM,the NDI values of 7.14%(1/14)and the ODI values of 71.43%(10/14)decreased,and the Viso value of 14.29%(2/14)increased in the aMCI group.The NDI value of right uncinate fasciculus,the ODI values of left superior longitudinal fasciculus,right hippocampus,left caudate nucleus,and left globus pallidus,and the Viso values of bilateral cingulum and left hippocampus significantly correlated with MMSE score.The NDI values of the splenium of corpus callosum,bilateral uncinate fasciculus and left amygdala,the ODI values of left superior longitudinal fasciculus,bilateral hippocampus,left parahippocampal gyrus,left caudate nucleus,left globus pallidus,left putamen and left thalamus,the Viso values of bilateral cingulum and left hippocampus significantly correlated with MoCA scores.Conclusion the decrease of neurite density was the main change of WM microstructure and the decrease of dendritic complexity was the main change of GM microstructure in the MCI stage.NODDI may reflect the clinical cognitive status of aMCI.NODDI combined with machine learning algorithm was expected to be a novel method for early diagnosis of MCI.