Trimodal imaging of human brain networks using simultaneous PET/MR/EEG

在过去的二十年中，数据驱动的处理方法已经确定了人类大脑的空间和时间上不同的功能网络，即使在所谓的“休息状态”中没有特定任务时，这些网络也是“活跃的”。这些静息态网络（RSN）中的许多在受试者之间非常相似，并且RSN的改变与包括癫痫和痴呆在内的脑部疾病有关。RSN甚至可以在非人类物种中识别，从而实现翻译方法并回答大脑功能的基本问题。然而，这些RSN的生理基础只是部分理解。绝大多数RSN研究基于使用血氧水平依赖（BOLD）对比度的功能MRI，即由血流动力学变化引起的最小MRI信号波动。因此，这种方法只能间接推断神经元的活动，并具有有限的时间分辨率。因此，从其他模态中识别RSN被认为是重要的一步。EEG和MEG探测神经元信号，但通常限于表面记录。尽管如此，使用适当的处理方法，RSN可以被识别为具有与fMRI中识别的网络相似的拓扑结构。[18 F]氟脱氧葡萄糖（FDG）-PET成像可以评估体内局部脑葡萄糖代谢（rCGM），并在可接受的空间但有限的时间分辨率下作为神经元能量消耗的替代品，因此，半定量静态[18 F]FDG-PET通常用于脑成像。在这个建议中，我们将获得完全同步的，动态的[18F]FDG-PET/fMRI/EEG数据在20个健康对照在休息状态，此外，使用一个简单的运动任务。我们将使用数据驱动的方法从所有三种模态中提取RSN，并具体分析RSN在模态和时间尺度之间的空间和时间关系。在最近的一项啮齿动物研究中，我们已经表明[18 F]FDG-PET和fMRI衍生的RSN是不相同的，更好地理解这些过程的生理学是至关重要的。我们现在将把这些方法和发现转化到人脑中，并通过并行EEG获取更直接的神经元信号来增加另一个时间和生理维度。利用校内资金，我们已经建立了必要的处理管道，并解决了该项目的技术先决条件。除了在健康志愿者中进行扫描外，我们还将获得一组接受相同成像模式的癫痫患者，以验证这种疾病中描述的RSN改变，并更好地定义[18 F]FDG-PET中常见的与致痫灶相关的rCGM降低的基础。该提案的结果将对整个大脑连接社区具有高度重要性，并将为大脑功能和不同成像模式下的大脑网络解释提供进一步的基本见解。

Over the last two decades data-driven processing methods have identified spatially and temporarily distinct functional networks of the human brain that are ‘active’ even at the absence of a specific task in the so-called ‘resting-state’. Many of these resting-state networks (RSN) are remarkably similar between subjects and alterations of RSNs have been associated with brain diseases including epilepsy and dementia. RSN could even be identified in non-human species enabling translational approaches and answering fundamental questions of brain functions. However, the physiological basis of these RSN is only partially understood. The vast majority of RSN studies is based on functional MRI using the blood oxygen level dependent (BOLD) contrast, i.e. minimal MRI signal fluctuations induced by hemodynamic changes. Thus, this approach can only indirectly infer neuronal activity and has limited temporal resolution. Therefore, identification of RSNs from other modalities is considered to be an important step. EEG and MEG probe a neuronal signal, but are usually limited to surface recordings. Still, using adequate processing methods, RSN could be identified that have a similar topology as the networks identified in fMRI. [18F]fluorodeoxyglucose (FDG)-PET imaging can assess regional cerebral glucose metabolism (rCGM) in-vivo and serves as a surrogate for neuronal energy consumption at acceptable spatial but limited temporal resolution, therefore, semi-quantitative static [18F]FDG-PET is generally used for brain imaging. Within this proposal we will acquire fully-simultaneous, dynamic [18F]FDG-PET/fMRI/EEG data in 20 healthy controls in the resting-state and, in addition, using a simple motor-task. We will extract RSN using data-driven methods from all three modalities and specifically analyze the spatial and temporal relation of RSN between the modalities and timescales. In a recent study in rodents we could already show that [18F]FDG-PET and fMRI derived RSNs are not identical and that a better understanding of the physiology of these processes is paramount. We will now translate the methods and findings to the human brain and add another temporal and physiological dimension by acquiring a more direct neuronal signal with parallel EEG. Using intramural funding, we have already established the necessary processing pipeline and solved the technical prerequisite for this project. In addition to scans in healthy volunteers, we will acquire a group of epilepsy patients undergoing the same imaging paradigm to validate RSN alterations described in this disease condition and better define the basis of the commonly observed decrease of rCGM in [18F]FDG-PET related to the epileptogenic focus. The results of this proposal will be of high importance for the whole brain connectivity community and will provide further, fundamental insights into brain functions and the interpretation of brain networks in the distinct imaging modalities.