Fu Guifang, Wan Nicholas J A, Baker Joseph M, Montgomery James W, Evans Julia L, Gillam Ronald B
Department of Mathematics and Statistics, Utah State University Logan, UT, USA.
Department of Psychology, Utah State University Logan, UT, USA.
Front Behav Neurosci. 2016 Jun 7;10:108. doi: 10.3389/fnbeh.2016.00108. eCollection 2016.
Functional near infrared spectroscopy (fNIRS) is a neuroimaging technology that enables investigators to indirectly monitor brain activity in vivo through relative changes in the concentration of oxygenated and deoxygenated hemoglobin. One of the key features of fNIRS is its superior temporal resolution, with dense measurements over very short periods of time (100 ms increments). Unfortunately, most statistical analysis approaches in the existing literature have not fully utilized the high temporal resolution of fNIRS. For example, many analysis procedures are based on linearity assumptions that only extract partial information, thereby neglecting the overall dynamic trends in fNIRS trajectories. The main goal of this article is to assess the ability of a functional data analysis (FDA) approach for detecting significant differences in hemodynamic responses recorded by fNIRS. Children with and without SLI wore two, 3 × 5 fNIRS caps situated over the bilateral parasylvian areas as they completed a language comprehension task. FDA was used to decompose the high dimensional hemodynamic curves into the mean function and a few eigenfunctions to represent the overall trend and variation structures over time. Compared to the most popular GLM, we did not assume any parametric structure and let the data speak for itself. This analysis identified significant differences between the case and control groups in the oxygenated hemodynamic mean trends in the bilateral inferior frontal and left inferior posterior parietal brain regions. We also detected significant group differences in the deoxygenated hemodynamic mean trends in the right inferior posterior parietal cortex and left temporal parietal junction. These findings, using dramatically different approaches, experimental designs, data sets, and foci, were consistent with several other reports, confirming group differences in the importance of these two areas for syntax comprehension. The proposed FDA was consistent with the temporal characteristics of fNIRS, thus providing an alternative methodology for fNIRS analyses.
功能近红外光谱技术（fNIRS）是一种神经成像技术，它使研究人员能够通过含氧血红蛋白和脱氧血红蛋白浓度的相对变化间接监测体内的大脑活动。fNIRS的一个关键特性是其卓越的时间分辨率，能够在非常短的时间内（以100毫秒为增量）进行密集测量。不幸的是，现有文献中的大多数统计分析方法尚未充分利用fNIRS的高时间分辨率。例如，许多分析程序基于线性假设，只提取部分信息，从而忽略了fNIRS轨迹中的整体动态趋势。本文的主要目标是评估功能数据分析（FDA）方法检测fNIRS记录的血流动力学反应显著差异的能力。患有和未患有特定语言障碍（SLI）的儿童在完成语言理解任务时，在双侧颞叶周围区域佩戴两顶3×5的fNIRS帽。FDA用于将高维血流动力学曲线分解为均值函数和一些特征函数，以表示随时间的整体趋势和变化结构。与最流行的广义线性模型（GLM）相比，我们没有假设任何参数结构，而是让数据自行说明问题。该分析确定了病例组和对照组在双侧额下回和左侧顶下后叶脑区含氧血流动力学平均趋势上的显著差异。我们还检测到右侧顶下后叶皮质和左侧颞顶交界区脱氧血流动力学平均趋势上的显著组间差异。这些使用截然不同的方法、实验设计、数据集和焦点得出的结果与其他几份报告一致，证实了这两个区域在句法理解重要性方面的组间差异。所提出的FDA与fNIRS的时间特征一致，从而为fNIRS分析提供了一种替代方法。
