静息态功能磁共振成像数据中几个常用指标的相关性

Correlation of several indexes in resting-state functional magnetic resonance imaging data

目的:基于功能磁共振成像(fMRI)技术,采用不同的指标来研究静息态功能数据,将局部一致性(ReHo)、分数低频振幅(fALFF)和Hurst指数进行对比研究,探求三者在大脑中的分布规律以及它们之间的内在联系。方法:对25例健康成人重复进行3次静息态fMRI扫描,利用MATLAB、DPARSF和REST软件进行数据处理以及计算ReHo、fALFF,并利用改进的周期图方法估计体素时间序列的Hurst指数,然后采用统计分析和回归分析方法进一步处理。结果:3个指标在全脑中分布基本一致,且在常见的静息态子网络中的取值一般都较大,有较好的稳定性。Hurst指数与ReHo成二次函数正相关,而Hurst指数与fALFF值成显著线性正相关。结论:ReHo、fALFF和Hurst指数具有区分不同的静息态子网络的潜力。3个指标两两之间都有很强的正相关性,分析数据时同时考虑三者有助于提高结果的准确性。

Objective Based on functional magnetic resonance imaging（fMRI） technology and resting- state fMRI data, to study on the distribution law of regional homogeneity（ReHo）, fractional amplitude of low frequency fluctuation（fALFF） and Hurst exponent in brain and the intrinsic link between them by comparatively analyzing ReHo, fALFF and Hurst exponent.Methods After 25 healthy adult receiving resting- state fMRI scan three times repeatedly, MATLAB, DPARSF and REST softwares were adopted for data processing, and computing ReHo and fALFF. The improved periodogram method was used to estimate Hurst exponent of time series in each voxel, and the statistical analysis and regression analysis were used for further processing. Results The distributions of ReHo, fALFF and Hurst exponent were almost consistent throughout the brain and their values in the common resting- state sub- network were usually larger and have better reliability. Hurst exponent was positively correlated with ReHo in quadratic function, and had a significant linear correlation with fALFF value. Conclusion ReHo, fALFF and Hurst exponent have the capacity to distinguish different resting-state sub-networks. The 3 indexes have a strong positive correlation between each other. Taking all of the 3 mdexes into account would improve the accuracy of analyzed results.