ANN generation according to a connection map of cultured network of living neurons on a dish

To elucidate higher brain function, it is essential to estimate the functional synaptic connectivity between neurons. For this purpose, it is a suitable method that analyzing responses of living neuronal network evoked by designed stimulation. In this study, we selected a dissociated culture of rat neuronal cells as an experimental material. The connection map of the cultured neuronal network was generated from spike trains recorded by using extracellular potential multisite recording system, which can record evoked and spontaneous action potentials in a neuronal network on a culture dish with 64 multi-electrodes arranged in lattice. We developed automatic estimating system for functional connections between neurons. The system detects electrical responses evoked by current stimulation, changing the stimulation electrode. By scanning every 64 electrodes as a stimulation site, the every relationship between a stimulation site and rest of 63 recording sites are estimated. It enables us to design a flexible stimulation schemes. The system outputs number of extracellular potentials in spontaneous and evoked activity recorded by multi-electrodes. Evoked responses were identified as a consequence of stimulation to a specific electrode. The connection matrix represents efficacy of connections between every possible combination of electrodes. Efficacy of connection was defined as mean rate of detected evoked spikes in each pair. According to the connection matrix, neuronal functional connections between one and other 63 electrodes were depicted into a map. We are also developing the system to generate a clone Artificial Neural Network (cANN) with the same connections as ones in a connection map of a living neuronal network. Our goal is developing cANN as a model of the target living neuronal network.