摘要
目的基于原始时间序列一阶导的相关分析构建功能连接(derivative functional connectivity,dFC),考察其检测精神分裂症(schizophrenia,SZ)患者脑功能异常的能力以及其对SZ的辅助诊断作用。方法基于一阶自回归模型构建仿真时间序列,比较功能连接(functional connectivity,FC)和dFC刻画时间序列相关性的异同。纳入91例SZ患者和91名健康对照(healthy controls,HC)进行全脑静息态功能磁共振扫描,计算116个脑区间的FC和dFC,分别针对FC和dFC进行逐条连接的组间比较,再利用多变量模式分析方法基于FC和dFC区分SZ和HC,以评估dFC在SZ辅助诊断方面的价值。结果仿真数据分析结果表明,相比于FC,dFC对时间序列的自相关性更不敏感,更能反映两个脑区神经活动时间序列之间的互相关关系。真实数据分析结果表明,SZ组和HC组间比较FC和dFC结果有较大重叠性,即49条连接的FC和dFC组间差异有统计学意义(P<0.05),也表现出明显的互补性,即36条连接仅FC显示组间差异有统计学差异(P<0.05),14条仅dFC显示组间差异有统计学意义(P<0.05)。FC与dFC区分SZ和HC的正确率分别为82.98%和82.40%,而两者融合后可进一步提高分类准确率至84.59%。结论dFC具有刻画脑功能连接信息的能力,与传统FC有互补性,两者联合能更全面地刻画SZ的脑异常连接模式,因此,dFC也可作为SZ辅助诊断的影像学生物标记物。
Objective To investigate the ability of functional connectivity constructed based on the correlation of the temporal derivatives of the original time series(i.e.,derivative functional connectivity)to detect brain abnormalities in patients with schizophrenia and its diagnostic value.Methods First,functional connectivity(FC)and derivative functional connectivity(dFC)were calculated based on simulated time series generated using an autoregressive model and then compared to characterize their similarities and differences.Second,FC and dFC were calculated based on real resting-state functional magnetic resonance imaging(rs-fMRI)data from 91 patients with schizophrenia(SZ)and 91 healthy controls(HC),and the differences in FCs and dFCs between SZ and HC were identified using univariate twosample t test.Finally,multivariate pattern analysis(MVPA)was used to distinguish SZ and HC based on whole-brain FC and dFC patterns,to assess dFC's diagnostic value for SZ.Results Simulated data showed that dFC was less sensitive to the autocorrelation of the time series compared to FC,indicating that dFC might better reflect the cross-correlation of neural activities in two brain regions.Real rs-fMRI data showed that FC and dFC also provided complementary information(36 connections showed significant between-group differences only for FC,and 14 showed significant differences only for dFC;P<0.05)although FC and dFC exhibited large overlap(both FC and dFC of 49 connections showed significant differences between SZ and HC;P<0.05).Furthermore,MVPA showed that both the FC pattern and dFC pattern could successfully differentiate SZ from HC(classification accuracies were 82.98%for FC and 82.40%for dFC),and the fusion of the two could further improve the classification accuracy to 84.59%.Conclusion dFC has the ability to measure brain functional connectivity and provides complementary information to conventional FC.Therefore,combination of FC and dFC can provide more comprehensive information about the brain abnormalities in SZ and thus dFC c
作者
周雪
宋颖超
郑明明
陈亚媛
秦文
梁猛
ZHOU Xue;SONG Yingchao;ZHENG Mingming;CHEN Yayuan;QIN Wen;LIANG Meng(School of Medical Technology,School of Medical Imaging,Tianjin Medical University,Tianjin 300203,China;Tianjin Key Laboratory of Functional Imaging,Tianjin 300052,China)
出处
《中国神经精神疾病杂志》
CAS
CSCD
北大核心
2023年第1期16-22,共7页
Chinese Journal of Nervous and Mental Diseases
基金
国家自然科学基金面上项目(编号:81971599)
国家重点研发计划基金(编号:2017YFC0909201)。
关键词
精神分裂症
磁共振成像
静息态功能磁共振成像
功能连接
多变量模式分析
一阶导数
Schizophrenia
Magnetic resonance imaging
Resting-state functional magnetic resonance imaging
Functional connectivity
Multivariate pattern analysis
First-order derivative
