Localization And Topology Mechanism Of "Learning Neurons" In Cultured Neuronal Networks Based On HD-MEA

In vitro cultured neurons have become a prevalent method to study the characteristics and mechanism of network-level learning.The information transmission of neurons must be based on the network to complete advanced cognitive functions such as learning and memory.How to locate the sparsely distributed clusters of "learning neurons" with the ability to code information cooperatively is essential for neural encoding and decoding.At the same time,it is of great significance to study the information processing mechanism of the neural system to study the dynamic changes of the functional network composed of "learning neurons" in information processing.In order to solve this problem,this paper proposed a method to locate "learning neurons" in cultivating neuron networks,studies the topological dynamics of "learning neurons" in information processing,and analyzes the learning mechanism of cultured neural networks.The multi-electrode array(MEA)has the characteristics of non-destructive recording,which can record long-term electrophysiological signals of neuronal networks,and is a common tool for network-level neuroinformatics research.This paper used a high-density MEA(HD-MEA)with complementary metal-oxidesemiconductor(CMOS)technology,and we used square wave current stimulation as a conditional stimulus to train the cultured neural network,and a method to locate the "learning neuron" in the cultured neural network was designed.The topological and dynamic characteristics of spontaneous and evoked firing of "learning neurons" in information processing were studied by using the methods of information theory and graph theory.At the same time,through the response pattern of "learning neurons",explored the influence of gamma-aminobutyric acid-A(GABAA)receptors on the topological dynamics of "learning neurons".In the study,the signals of the cultured neuron network were preprocessed using conventional neuro-electrophysiological analysis methods,and the network spike was extracted.The electrophysiological signals of 4096 electrodes were linearly fitted multiple times.It was determined that the spatial position of neurons was "learning neurons" cluster with the gradually increased of response stimulus ratio and the gradually decreased of response time.Data analysis shows that the average firing rate of "learning neurons" within 1 hour after learning was 242% higher than that before learning,and it exhibits network plasticity and produced a stable response to electrical stimulation after learning 4 hours.In the process of stimulation,the network density,network efficiency and small world of the functional network constructed by "learning neurons" increased by 83.3%,133.4% and 230%.The spontaneous analysis results show that the correlation of "learning neurons" increased by 341% during the stimulation process,while the network density increased by 37.5%,and the global efficiency increased by 46.5%,but the small-world always remained above 1.The modularity of the network increased gradually during the stimulation process.The Euclidean distance statistics of the burst initiation point and the “learning neuron”found that 99.7% of the burst start initiation was within 162 μm of the “learning neuron”.The inhibition of bicuculline by GABAA receptors caused significant changes in the dynamics of spontaneous activities,including the following features: the number of bursts per unit time increased by 125%,the duration of bursts increased by 25.3%,and the network synchronization increased by 34.7 %,the correlation increased by 29.1%,the network density increased by 244.5%,the global efficiency increased by 193.5%,and the small world increased by 72%.This paper has studied the topological dynamics of "learning neurons" in spontaneous and evoked activities,and found that "learning neurons" were the key factor connecting spontaneous and evoked activities.At the same time,it is found that the cause of network bursts after learning was closely related to "learning neurons",which puts forward a new view on the cause of network bursts after learning.This paper provides a new idea for revealing the group activity coding law of neural network information and the neural information processing mechanism.