Study On The Nonlinear Dynamics Of Neuronal Networks Cultured In Vitro

Understanding the information processing mechanism of nervous system presents agreat challenge to investigators in neuroscience. One way to understand informationprocessing mechanism is to study the dynamics of electrical signals which play animportant role in neural coding. Using multielectrode array system, we recorded thespontaneous activity of cultured hippocampal neuronal networks for a long-term, andincorporated several nonlinear methods to study the characteristics of dynamical behaviorat the mesoscopic level.The electrical signals of cultured hippocampal neuronal networks were preprocessedfor further study. To detect bursting events from spike train which was recorded by oneelectrode, a new algorithm based on recurrence quantification analysis was established.Experimental results indicated that the average validity for the burst detection wasachieved about92.89±2.64%. To detect synchronous burst event from population activitywhich was recorded by60electrodes, this paper improved the existing algorithm withself-adaptive threshold, which increased the average validity to91.17±3.11%. Then theinherent stability of electrical signal was studied, and the results indicated that nosignificant change occurred in the characteristics during1to2h.To understand the complex dynamics exhibited by cultured hippocampal neuronalnetworks, nonlinear methods have been attempted to analyze the electrical signals. Thisstudy calculated the largest Lyapunov exponent, measured fractal exponent, and appliedapproximate entropy to local dynamics and global dynamics. By comparing theconsistence, the robustness to noise and the ability to distinguish network state, we foundthe first clear evidence that the largest Lyapunov exponent was well-suited for studyingthe chaotic behavior of local dynamics, and fractal exponent for the fractal behavior ofglobal dynamics.With10long-term cultured hippocampal neuronal networks, we applied the largestLyapunov exponent to study the development related changes in the chaotic behavior oflocal dynamics. The results demonstrated that chaotic behavior emerged in the early andmiddle stage of neural development (18.98±2.30%to41.35±3.64%). The temporal evolution of chatic behavior showed an emergent and periodic transition between chaosand bifurcation, and the road to chaos is intermittency. The spatial distribution of chaoticbehavior presented network-wide synchronization, as well as a significant positivecorrelation between the degree of chaos and the number of active node in the entirenetworkThe alternation of global dynamics during neural development was measurd byfractal exponent. The results indicated that the fractal dynamics of cultured hippocampalneuronal networks took the form of an inverted U-shaped function across the life-span. Toreveal the mechanisms of generation and evolution of fractal dynamics in developingneuronal networks, further study about the relationship between fractal exponent andfunctional organization has been carried out. The results indicated that the change offractal dynamcis referred to alternations in both the number of functional components andthe small world topology of functional organization.This paper studied the nonlinear dynamics of neuronal networks at mesoscopic level,which was helpful to extract information hidden in electrical signals and provided newinsight to neural information processing mechanism.