The dynamic connectome: towards large-scale 3D reconstruction of brain activity in real-time

What does large-scale connectivity tell us about whole-brain activity and neural circuits? In this work, we present a virtual reality based large-scale dynamic simulation for 3D reconstruction of whole-brain activity over the cortical connectome in real-time. Using DTI structural connectivity data from [1] we built an interactive 3D visualization of the human connectome network in an immersive virtual reality environment (Figure 1A) using the Unity 3D gaming engine. Further, the virtual reality brain network in Unity is coupled to a real-time neuronal simulator, iqr [2]. As we see, coupling structural connectivity data with detailed enough neuronal population dynamics is sufficient in predicting functional correlations and large-scale activity patterns. We model neuronal dynamics by a linear-threshold filter (as work in progress, we are currently implementing population dynamics from mean-field models [3]). Each population module is stochastic, having Gaussian noise. The user can stimulate any region or simultaneous regions of the network with external input currents. The simulation then reconstructs reverberating neural activity propagating throughout the network in real-time. As an explicit example, we stimulate the superior parietal areas and observe causal activity propagation in the parietal lobe, indicative of visuo-motor integration (Figure 1B). This is a first step to simulating and mapping large-scale brain activity in real-time.