Center for Systems and Translational Brain Sciences, Institute of Human Complexity and Systems Science, Yonsei University, Seoul, Republic of Korea; Department of Nuclear Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea.
Center for Systems and Translational Brain Sciences, Institute of Human Complexity and Systems Science, Yonsei University, Seoul, Republic of Korea; Department of Nuclear Medicine, Yonsei University College of Medicine, Seoul, Republic of Korea; Institute of Neuroscience and Medicine, Brain and Behaviour (INM-7), Research Centre Jülich, Germany.
Neuroimage. 2020 Jun;213:116755. doi: 10.1016/j.neuroimage.2020.116755. Epub 2020 Mar 19.
The aim of this paper is to present a dynamic causal modeling (DCM) framework for hippocampal activity measured via voltage-sensitive dye imaging (VSDI). We propose a DCM model of the hippocampus that summarizes interactions between the hilus, CA3 and CA1 regions. The activity of each region is governed via a neuronal mass model with two inhibitory and one/two excitatory neuronal populations, which can be linked to measurement VSDI by scaling neuronal activity. To optimize the model structure for the hippocampus, we propose two Bayesian schemes: Bayesian hyperparameter optimization to estimate the unknown electrophysiological properties necessary for constructing a mesoscopic hippocampus model; and Bayesian model reduction to determine the parameterization of neural properties, and to test and include potential connections (morphologically inferred without direct evidence yet) in the model by evaluating group-level model evidence. The proposed method was applied to model spatiotemporal patterns of accumulative responses to consecutive stimuli in separate groups of wild-type mice and epileptic aristaless-related homeobox gene (Arx) conditional knock-out mutant mice (Arx;Dlx5/6) in order to identify group differences in the effective connectivity within the hippocampus. The causal role of each group-differing connectivity in generating mutant-like responses was further tested. The group-level analysis identified altered intra- and inter-regional effective connectivity, some of which are crucial for explaining mutant-like responses. The modelling results for the hippocampal activity suggest the plausibility of the proposed mesoscopic hippocampus model and the usefulness of utilizing the Bayesian framework for model construction in the mesoscale modeling of neural interactions using DCM.
本文旨在提出一个基于电压敏感染料成像（VSDI）测量的海马体活动的动态因果建模（DCM）框架。我们提出了一个海马体的 DCM 模型，该模型总结了神经细胞体模型中的各个区域（包括齿状回、CA3 和 CA1 区域）之间的相互作用。每个区域的活动由具有两个抑制性和一个/两个兴奋性神经元群体的神经元群体模型来控制，通过对神经元活动进行缩放，可以将其与 VSDI 测量联系起来。为了优化海马体模型的结构，我们提出了两种贝叶斯方案：贝叶斯超参数优化，用于估计构建介观海马体模型所需的未知电生理特性；以及贝叶斯模型降阶，用于确定神经特性的参数化，并通过评估组水平模型证据来测试和包括模型中的潜在连接（没有直接证据的形态学推断）。该方法应用于模型空间时间模式的累积反应，对野生型小鼠和癫痫性 aristaless 相关同源盒基因（Arx）条件敲除突变体小鼠（Arx;Dlx5/6）的连续刺激，以确定海马体中有效连接的组间差异。进一步测试了每个组间差异连接在产生突变样反应中的因果作用。组水平分析确定了改变的内部和区域间有效连接，其中一些对于解释突变样反应至关重要。对海马体活动的建模结果表明，所提出的介观海马体模型具有合理性，并且利用贝叶斯框架进行模型构建在利用 DCM 进行神经相互作用的介观建模中是有用的。
