Faculty of Human Sciences, Institute III, Department of Sport Science, Otto von Guericke University, Zschokkestraße 32, 39104 Magdeburg, Germany.
Faculty of Human Sciences, Institute III, Department of Sport Science, Otto von Guericke University, Zschokkestraße 32, 39104 Magdeburg, Germany; Department of Neurology, Max Planck Institute for Human Cognitive and Brain Sciences, Stephanstraße 1a, 04103 Leipzig, Germany.
Neuroimage. 2022 Aug 1;256:119249. doi: 10.1016/j.neuroimage.2022.119249. Epub 2022 Apr 27.
Multiparameter mapping (MPM) is a quantitative MRI protocol that is promising for studying microstructural brain changes in vivo with high specificity. Reliability values are an important prior knowledge for efficient study design and facilitating replicable findings in development, aging and neuroplasticity research. To explore longitudinal reliability of MPM we acquired the protocol in 31 healthy young subjects twice over a rescan interval of 4 weeks. We assessed the within-subject coefficient of variation (WCV), the between-subject coefficient of variation (BCV), and the intraclass correlation coefficient (ICC). Using these metrics, we investigated the reliability of (semi-) quantitative magnetization transfer saturation (MT), proton density (PD), transversal relaxation (R2*) and longitudinal relaxation (R1). To increase relevance for explorative studies in development and training-induced plasticity, we assess reliability both on local voxel- as well as ROI-level. Finally, we disentangle contributions and interplay of within- and between-subject variability to ICC and assess the optimal degree of spatial smoothing applied to the data. We reveal evidence that voxelwise ICC reliability of MPMs is moderate to good with median values in cortex (subcortical GM): MT: 0.789 (0.447) PD: 0.553 (0.264) R1: 0.555 (0.369) R2*: 0.624 (0.477). The Gaussian smoothing kernel of 2 to 4 mm FWHM resulted in optimal reproducibility. We discuss these findings in the context of longitudinal intervention studies and the application to research designs in neuroimaging field.
多参数映射（MPM）是一种定量 MRI 协议，具有很高的特异性，有望在体内研究微观结构的大脑变化。可靠性值是高效研究设计的重要先验知识，有助于在发育、衰老和神经可塑性研究中获得可复制的发现。为了探索 MPM 的纵向可靠性，我们在 4 周的重扫间隔内对 31 名健康年轻受试者进行了两次协议采集。我们评估了个体内变异系数（WCV）、个体间变异系数（BCV）和组内相关系数（ICC）。使用这些指标，我们研究了（半）定量磁化转移饱和（MT）、质子密度（PD）、横向弛豫（R2*）和纵向弛豫（R1）的可靠性。为了增加在发育和训练诱导可塑性的探索性研究中的相关性，我们在局部体素和 ROI 水平上评估可靠性。最后，我们将个体内和个体间变异性的贡献和相互作用分解开来，并评估应用于数据的最佳空间平滑度程度。我们的研究结果表明，MPM 的体素内 ICC 可靠性为中等至良好，皮质（皮质下 GM）的中位数值为：MT：0.789（0.447）PD：0.553（0.264）R1：0.555（0.369）R2*：0.624（0.477）。高斯平滑核的 FWHM 为 2 至 4 毫米时，可获得最佳的重现性。我们将这些发现置于纵向干预研究的背景下进行讨论，并将其应用于神经影像学领域的研究设计。
