Silva João Paulo Santos, Mônaco Luciana da Mata, Paschoal André Monteiro, Oliveira Ícaro Agenor Ferreira de, Leoni Renata Ferranti
Inbrain Lab, Department of Physics, FFCLRP, University of Sao Paulo, Ribeirao Preto, Brazil.
Inbrain Lab, Department of Physics, FFCLRP, University of Sao Paulo, Ribeirao Preto, Brazil.
Magn Reson Imaging. 2018 Sep;51:151-157. doi: 10.1016/j.mri.2018.05.006. Epub 2018 May 16.
Arterial spin labeling (ASL) is an established magnetic resonance imaging (MRI) technique that is finding broader applications in functional studies of the healthy and diseased brain. To promote improvement in cerebral blood flow (CBF) signal specificity, many algorithms and imaging procedures, such as subtraction methods, were proposed to eliminate or, at least, minimize noise sources. Therefore, this study addressed the main considerations of how CBF functional connectivity (FC) is changed, regarding resting brain network (RBN) identification and correlations between regions of interest (ROI), by different subtraction methods and removal of residual motion artifacts and global signal fluctuations (RMAGSF).
Twenty young healthy participants (13 M/7F, mean age = 25 ± 3 years) underwent an MRI protocol with a pseudo-continuous ASL (pCASL) sequence. Perfusion-based images were obtained using simple, sinc and running subtraction. RMAGSF removal was applied to all CBF time series. Independent Component Analysis (ICA) was used for RBN identification, while Pearson' correlation was performed for ROI-based FC analysis.
Temporal signal-to-noise ratio (tSNR) was higher in CBF maps obtained by sinc subtraction, although RMAGSF removal had a significant effect on maps obtained with simple and running subtractions. Neither the subtraction method nor the RMAGSF removal directly affected the identification of RBNs. However, the number of correlated and anti-correlated voxels varied for different subtraction and filtering methods. In an ROI-to-ROI level, changes were prominent in FC values and their statistical significance.
Our study showed that both RMAGSF filtering and subtraction method might influence resting-state FC results, especially in an ROI level, consequently affecting FC analysis and its interpretation. Taking our results and the whole discussion together, we understand that for an exploratory assessment of the brain, one could avoid removing RMAGSF to not bias FC measures, but could use sinc subtraction to minimize low-frequency contamination. However, CBF signal specificity and frequency range for filtering purposes still need to be assessed in future studies.
动脉自旋标记（ASL）是一种成熟的磁共振成像（MRI）技术，在健康和患病大脑的功能研究中有着越来越广泛的应用。为了提高脑血流量（CBF）信号的特异性，人们提出了许多算法和成像程序，如减法方法，以消除或至少最小化噪声源。因此，本研究探讨了不同减法方法以及去除残余运动伪影和全局信号波动（RMAGSF）对静息脑网络（RBN）识别和感兴趣区域（ROI）之间相关性方面CBF功能连接（FC）变化的主要影响因素。
20名年轻健康参与者（13名男性/7名女性，平均年龄 = 25 ± 3岁）接受了包含伪连续ASL（pCASL）序列的MRI检查方案。使用简单减法、正弦减法和连续减法获取基于灌注的图像。对所有CBF时间序列应用RMAGSF去除。独立成分分析（ICA）用于RBN识别，而基于ROI的FC分析采用Pearson相关性分析。
通过正弦减法获得的CBF图的时间信噪比（tSNR）更高，尽管RMAGSF去除对通过简单减法和连续减法获得的图有显著影响。减法方法和RMAGSF去除均未直接影响RBN的识别。然而，不同减法和滤波方法的相关和反相关体素数量有所不同。在ROI到ROI水平上，FC值及其统计显著性变化显著。
我们的研究表明，RMAGSF滤波和减法方法都可能影响静息状态FC结果，尤其是在ROI水平，从而影响FC分析及其解释。综合我们的结果和整个讨论，我们明白对于大脑的探索性评估，可以避免去除RMAGSF以免对FC测量产生偏差，但可以使用正弦减法来最小化低频污染。然而，CBF信号特异性和滤波目的的频率范围仍需在未来研究中进行评估。
