Multivariate Dynamical Systems Methods for Identifying Causal Interactions in fMR

描述（由申请人提供）：认知信息处理依赖于分布的大脑区域之间的动态相互作用。在过去的十年中，功能磁共振成像（fMRI）已经成为研究人类大脑功能的有力工具。尽管fMRI研究主要集中在识别在执行认知任务时被激活的大脑区域，但越来越多的人认为，认知功能是多个大脑区域之间动态、情境依赖、因果相互作用的结果。因此，设计和验证研究这种相互作用的方法具有更大的意义。尽管需求不断增长，但目前识别fMRI数据中因果相互作用的方法的准确性仍然知之甚少。本提案的总体目标是通过开发和测试新的算法和软件来识别分布的大脑区域之间依赖于上下文的因果相互作用，从而解决功能磁共振成像的关键需求。我们将首先开发和验证基于多元动力系统（MDS）框架的新方法，该框架克服了现有方法的几个限制。然后，我们将在模拟和真实fMRI数据上比较我们的新方法与其他方法的性能。这些研究的重要贡献包括：(1)开发了新的多元状态空间方法来估计大脑区域之间的因果相互作用；(2)首次和最详细地评估MDS以及使用模拟和实验fMRI数据的其他有效连接方法。总之，这些研究将导致新的和改进的工具来分析功能性脑连接使用fMRI。更广泛地说，我们提出的方法将有助于推进对人类认知功能动力学基础的了解，并将为研究神经发育、精神和神经疾病（如自闭症、精神分裂症和帕金森病）提供新的工具。拟议的研究与美国国立卫生研究院生物医学计算科学与技术探索创新计划公告（PA 09-219）的使命高度相关，该计划旨在鼓励开发用于脑成像的创新先进计算工具。

DESCRIPTION (provided by applicant): Cognitive information processing depends on dynamical interactions between distributed brain areas. In the past decade, functional magnetic resonance imaging (fMRI) has emerged as a powerful tool for investigating human brain function. Although fMRI research has primarily focused on identifying brain regions that are activated during performance of cognitive tasks, there is growing consensus that cognitive functions emerge as a result of dynamic, context-dependent, causal interactions between multiple brain areas. Devising and validating methods for investigating such interactions has therefore taken added significance. Despite the growing need, the accuracy of current methods for identifying causal interactions in fMRI data remain poorly understood. The overall goal of this proposal is to address a critical need in fMRI by developing and testing new algorithms and software for identifying context-dependent causal interactions between distributed brain regions. We will first develop and validate novel methods based on a Multivariate Dynamical Systems (MDS) framework that overcomes several limitations of existing methods. We will then compare the performance of our new methods with other methods on both simulated and real fMRI data. Important contributions of these proposed studies include (1) development of novel multivariate state space methods for estimating causal interactions between brain regions and (2) first and most detailed evaluation of not only MDS but also other effective connectivity methods using both simulated and experimental fMRI data. Together, these studies will lead to new and improved tools for analyzing functional brain connectivity using fMRI. More generally, our proposed methods will help to advance knowledge of the dynamical basis of human cognitive function and will provide new tools for investigating neurodevelopmental, psychiatric and neurological disorders such as autism, schizophrenia and Parkinson's disease. The proposed studies are highly relevant to the mission of the NIH Exploratory Innovations in Biomedical Computational Science and Technology Program Announcement (PA 09-219), which seeks to encourage development of innovative advanced computational tools for brain imaging. PUBLIC HEALTH RELEVANCE: In the past decade, functional magnetic resonance imaging (fMRI) has emerged as a powerful tool for investigating human brain function and dysfunction. Although fMRI studies of brain function have primarily focused on identifying brain regions that are activated during performance of perceptual or cognitive tasks, there is growing consensus that cognitive functions emerge as a result of dynamic context-dependent interactions between multiple brain areas. Developing new methods for investigating causal interactions in fMRI data has therefore taken added significance; the overall goal of this proposal is to address this critical need by developing new methods for studying causal interactions and brain connectivity between distributed brain regions during cognition.