Greene Clint, Cieslak Matt, Grafton Scott T
Signal Compression Lab, Department of Electrical and Computer Engineering, University of California, Santa Barbara, CA, USA.
Action Lab, Department of Psychological and Brain Sciences, University of California, Santa Barbara, CA, USA.
Netw Neurosci. 2018 Sep 1;2(3):362-380. doi: 10.1162/netn_a_00035. eCollection 2018.
To facilitate the comparison of white matter morphologic connectivity across target populations, it is invaluable to map the data to a standardized neuroanatomical space. Here, we evaluated direct streamline normalization (DSN), where the warping was applied directly to the streamlines, with two publically available approaches that spatially normalize the diffusion data and then reconstruct the streamlines. Prior work has shown that streamlines generated after normalization from reoriented diffusion data do not reliably match the streamlines generated in native space. To test the impact of these different normalization methods on quantitative tractography measures, we compared the reproducibility of the resulting normalized connectivity matrices and network metrics with those originally obtained in native space. The two methods that reconstruct streamlines after normalization led to significant differences in network metrics with large to huge standardized effect sizes, reflecting a dramatic alteration of the same subject's native connectivity. In contrast, after normalizing with DSN we found no significant difference in network metrics compared with native space with only very small-to-small standardized effect sizes. DSN readily outperformed the other methods at preserving native space connectivity and introduced novel opportunities to define connectome networks without relying on gray matter parcellations.
为便于比较不同目标人群的白质形态学连通性，将数据映射到标准化神经解剖空间非常重要。在此，我们评估了直接流线归一化（DSN），即将变形直接应用于流线，并与两种公开可用的方法进行了比较，这两种方法先对扩散数据进行空间归一化，然后重建流线。先前的研究表明，从重新定向的扩散数据归一化后生成的流线与在原始空间中生成的流线不能可靠匹配。为测试这些不同归一化方法对定量纤维束成像测量的影响，我们将所得归一化连通性矩阵和网络指标的可重复性与最初在原始空间中获得的结果进行了比较。在归一化后重建流线的两种方法导致网络指标出现显著差异，标准化效应大小从大到巨大，反映了同一受试者原始连通性的巨大改变。相比之下，在用DSN归一化后，我们发现与原始空间相比，网络指标没有显著差异，标准化效应大小仅为非常小到小。在保留原始空间连通性方面，DSN明显优于其他方法，并引入了无需依赖灰质分割来定义连接组网络的新机会。
