On the estimation of brain signal entropy from sparse neuroimaging data.
On the estimation of brain signal entropy from sparse neuroimaging data.
复制标题
DOI：
10.1038/srep23073
复制
发表时间：
2016-03-29
影响因子：
4.6
通讯作者：
Werkle-Bergner M
中科院分区：
文献类型：
--
作者：
Grandy TH;Garrett DD;Schmiedek F;Werkle-Bergner M
多尺度熵（MSE）最近被确立为一个有前途的工具，用于分析神经信号的时刻到时刻的变化。吸引人的是，MSE提供了一个衡量大脑在多个时间尺度上运行的神经操作的可预测性的指标。在某些类别的神经信号的MSE的应用程序的一个重要限制是MSE的明显依赖于长时间序列。然而，MSE计算中的这种稀疏数据限制可以潜在地经由跨不一定连续采集的较短时间序列的MSE估计来克服（例如，在fMRI块设计中）。在本研究中，使用模拟，EEG和fMRI数据，我们研究了MSE估计的准确性和精确度对每段数据点数量和数据段总数的依赖性。正如假设的那样，不连续节段的MSE估计是非常准确和精确的，尽管节段长度。我们的方法的一个关键的进步是，它允许计算MSE规模以前无法从本地段长度。因此，我们的研究结果可能会允许更广泛的应用范围内的MSE时，衡量的时刻到时刻的动态稀疏和/或不连续的神经生理数据典型的许多现代认知神经科学研究设计。
Multi-scale entropy (MSE) has been recently established as a promising tool for the analysis of the moment-to-moment variability of neural signals. Appealingly, MSE provides a measure of the predictability of neural operations across the multiple time scales on which the brain operates. An important limitation in the application of the MSE to some classes of neural signals is MSE’s apparent reliance on long time series. However, this sparse-data limitation in MSE computation could potentially be overcome via MSE estimation across shorter time series that are not necessarily acquired continuously (e.g., in fMRI block-designs). In the present study, using simulated, EEG, and fMRI data, we examined the dependence of the accuracy and precision of MSE estimates on the number of data points per segment and the total number of data segments. As hypothesized, MSE estimation across discontinuous segments was comparably accurate and precise, despite segment length. A key advance of our approach is that it allows the calculation of MSE scales not previously accessible from the native segment lengths. Consequently, our results may permit a far broader range of applications of MSE when gauging moment-to-moment dynamics in sparse and/or discontinuous neurophysiological data typical of many modern cognitive neuroscience study designs.