PRAGMA: Interactively Constructing Functional Brain Parcellations
PRAGMA: Interactively Constructing Functional Brain Parcellations
复制标题
DOI：
10.1109/vis47514.2020.00016
复制
发表时间：
2020-09
期刊：
影响因子：
--
通讯作者：
Roza G. Bayrak;Nhung Hoang;Colin B. Hansen;Catie Chang;M. Berger
中科院分区：
文献类型：
--
作者：
Roza G. Bayrak;Nhung Hoang;Colin B. Hansen;Catie Chang;M. Berger
神经影像学研究的一个突出目标是绘制人类大脑，以识别和描绘功能有意义的区域，并阐明其在认知行为中的作用。这些大脑区域通常由地图集表示，这些地图集捕捉了大量人群的一般趋势。尽管对神经影像学专家来说是不可或缺的，但群体水平的地图集并不能捕捉功能组织的个体差异。在这项工作中，我们提出了一个交互式的可视化方法，PRAGMA，它允许领域专家从建立的地图集获得扫描特定的包裹。PRAGMA的特点是用户驱动的，层次聚类方案，用于定义时间相关的包裹在不同的粒度。可视化设计支持用户决定如何执行聚类，即何时展开、折叠或合并地块。这是通过一组链接和协调的视图来实现的，这些视图用于理解用户的当前层次结构、评估集群内的变化以及将包裹与已建立的地图集相关联。我们通过与四位神经影像领域专家的用户研究评估了PRAGMA的有效性，我们的结果表明，PRAGMA显示出探索个性化和特定状态的大脑包裹的潜力，并提供了有趣的功能性大脑网络的见解。
A prominent goal of neuroimaging studies is mapping the human brain, in order to identify and delineate functionally-meaningful regions and elucidate their roles in cognitive behaviors. These brain regions are typically represented by atlases that capture general trends over large populations. Despite being indispensable to neuroimaging experts, population-level atlases do not capture individual differences in functional organization. In this work, we present an interactive visualization method, PRAGMA, that allows domain experts to derive scan-specific parcellations from established atlases. PRAGMA features a user-driven, hierarchical clustering scheme for defining temporally correlated parcels in varying granularity. The visualization design supports the user in making decisions on how to perform clustering, namely when to expand, collapse, or merge parcels. This is accomplished through a set of linked and coordinated views for understanding the user’s current hierarchy, assessing intra-cluster variation, and relating parcellations to an established atlas. We assess the effectiveness of PRAGMA through a user study with four neuroimaging domain experts, where our results show that PRAGMA shows the potential to enable exploration of individualized and state-specific brain parcellations and to offer interesting insights into functional brain networks.