Jakobsen Estrid, Böttger Joachim, Bellec Pierre, Geyer Stefan, Rübsamen Rudolf, Petrides Michael, Margulies Daniel S
Max Planck Institute for Human Cognitive and Brain Sciences, Stephanstrasse 1A, 04103, Leipzig, Germany.
Centre de recherche de l'institut de Gériatrie de Montréal, Montreal, QC, Canada.
Eur J Neurosci. 2016 Feb;43(4):561-71. doi: 10.1111/ejn.13140.
Broca's region is composed of two adjacent cytoarchitectonic areas, 44 and 45, which have distinct connectivity to superior temporal and inferior parietal regions in both macaque monkeys and humans. The current study aimed to make use of prior knowledge of sulcal anatomy and resting-state functional connectivity, together with a novel visualization technique, to manually parcellate areas 44 and 45 in individual brains in vivo. One hundred and one resting-state functional magnetic resonance imaging datasets from the Human Connectome Project were used. Left-hemisphere surface-based correlation matrices were computed and visualized in brainGL. By observation of differences in the connectivity patterns of neighbouring nodes, areas 44 and 45 were manually parcellated in individual brains, and then compared at the group-level. Additionally, the manual labelling approach was compared with parcellation results based on several data-driven clustering techniques. Areas 44 and 45 could be clearly distinguished from each other in all individuals, and the manual segmentation method showed high test-retest reliability. Group-level probability maps of areas 44 and 45 showed spatial consistency across individuals, and corresponded well to cytoarchitectonic probability maps. Group-level connectivity maps were consistent with previous studies showing distinct connectivity patterns of areas 44 and 45. Data-driven parcellation techniques produced clusters with varying degrees of spatial overlap with the manual labels, indicating the need for further investigation and validation of machine learning cortical segmentation approaches. The current study provides a reliable method for individual-level cortical parcellation that could be applied to regions distinguishable by even the most subtle differences in patterns of functional connectivity.
布洛卡区由两个相邻的细胞构筑区44区和45区组成，在猕猴和人类中，这两个区域与颞上区和顶下区有着不同的连接。本研究旨在利用脑沟解剖学和静息态功能连接的先验知识，结合一种新颖的可视化技术，在活体个体大脑中手动划分44区和45区。使用了来自人类连接体项目的101个静息态功能磁共振成像数据集。计算左半球基于表面的相关矩阵，并在brainGL中进行可视化。通过观察相邻节点连接模式的差异，在个体大脑中手动划分44区和45区，然后在组水平上进行比较。此外，将手动标记方法与基于几种数据驱动聚类技术的划分结果进行了比较。在所有个体中，44区和45区都能清晰区分，且手动分割方法显示出较高的重测信度。44区和45区的组水平概率图在个体间显示出空间一致性，并且与细胞构筑概率图吻合良好。组水平连接图与先前显示44区和45区不同连接模式的研究一致。数据驱动的划分技术产生的聚类与手动标记有不同程度的空间重叠，这表明需要进一步研究和验证机器学习皮质分割方法。本研究提供了一种可靠的个体水平皮质划分方法，该方法可应用于即使在功能连接模式上存在最细微差异也能区分的区域。
