Data from: Low-frequency fluctuations of the resting brain: high magnitude does not equal high reliability

The amplitude of low-frequency fluctuation (ALFF) measures low-frequency oscillations of the blood-oxygen-level-dependent signal, characterizing local spontaneous activity during the resting state. ALFF is a commonly used measure for resting-state functional magnetic resonance imaging (rs-fMRI) in numerous basic and clinical neuroscience studies. Using a test-retest rs-fMRI dataset consisting of 21 healthy subjects and three repetitive scans, we found that several key brain regions with high ALFF intensities (or magnitude) had poor reliability. Such regions included the posterior cingulate cortex, the medial prefrontal cortex in the default mode network, parts of the right and left thalami, and the primary visual and motor cortices. The above finding was robust with regard to different sample sizes (number of subjects), different scanning parameters (repetition time) and variations of test-retest intervals (i.e., intra-scan, intra-session, and inter-session reliability), as well as with different scanners. Moreover, the qualitative, map-wise results were validated further with a region-of-interest-based quantitative analysis using “canonical” coordinates as reported previously. Therefore, we suggest that the reliability assessments be incorporated in future ALFF studies, especially for the brain regions with a large ALFF magnitude as listed in our paper. Splitting single data into several segments and assessing within-scan “test-retest” reliability is an acceptable alternative if no “real” test-retest datasets are available. Such evaluations might become more necessary if the data are collected with clinical scanners whose performance is not as good as those that are used for scientific research purposes and are better maintained because the lower signal-to-noise ratio may further dampen ALFF reliability.低频振幅（Amplitude of Low-Frequency Fluctuation, ALFF）用于量化血氧水平依赖信号（blood-oxygen-level-dependent signal）的低频振荡特征，用以刻画静息态下脑局部的自发神经活动。ALFF是静息态功能磁共振成像（Resting-State Functional Magnetic Resonance Imaging, rs-fMRI）领域在诸多基础与临床神经科学研究中广泛应用的分析指标。本研究采用一项包含21名健康受试者、共3次重复扫描的重测rs-fMRI数据集开展分析，结果发现多个ALFF强度（幅值）较高的关键脑区存在较差的重测信度。此类脑区涵盖默认模式网络（Default Mode Network, DMN）中的后扣带回皮层、内侧前额叶皮层，双侧丘脑的部分区域，以及初级视觉皮层与初级运动皮层。上述研究结论在不同样本量（受试者数量）、不同扫描参数（重复时间，Repetition Time, TR）、不同重测间隔（即扫描内、会话内及会话间信度），以及不同扫描设备的条件下均表现出稳健性。此外，本研究通过采用既往报道的“标准（canonical）”坐标开展基于感兴趣区（Region of Interest, ROI）的定量分析，进一步验证了基于脑图的定性分析结果。据此，我们建议在未来的ALFF相关研究中纳入信度评估环节，尤其针对本文所列的ALFF幅值较高的脑区。若无法获取“真实”的重测数据集，可将单份数据拆分为若干分段，通过评估扫描内的“重测”信度作为可行替代方案。若使用临床扫描设备采集数据——此类设备的性能不及科研专用且维护更完善的设备，其较低的信噪比（Signal-to-Noise Ratio, SNR）可能进一步削弱ALFF的重测信度——则此类信度评估的必要性将显著提升。