BiœmuS: A new tool for neurological disorders studies through real-time emulation and hybridization using biomimetic Spiking Neural Network Article Swipe
YOU?
·
· 2023
· Open Access
·
· DOI: https://doi.org/10.21203/rs.3.rs-3191285/v1
· OA: W4386770001
<title>Abstract</title> Characterization and modeling of biological neural networks is a field opening to major advances in our understanding of the mechanisms governing the functioning of the brain and the different pathologies that can affect it. Recent researches in bioelectronics and neuromorphic engineering lead to the design of the new generation of neuroprosthesis. Here we show a novel real-time, biomimetic and energy-efficient neural network for bio-hybrid experiments and parallel emulation. This novel system is used to investigate and reproduce neural network dynamics. The setup is running on a digital platform using a System on Chip (SoC) featuring both Programmable Logic (PL) and processors in a Processing System (PS) part. The FPGA part is computing the biomimetic and real-time electrical activities of Hodgkin-Huxley neural network while the processors handle monitoring and communication. New methods of resource and power optimization has been applied to the FPGA to allow detailed neuron modeling with synapses showing short term plasticity. The system is validated by comparison with biological data and model. We also demonstrate the feasibility of bio-hybrid experiments with different bio-physical interface and different biological cells. The complete setup achieves communication with a fully flexible real-time device thus constituting a step towards neuromorphic-based neuroprosthesis for bioelectrical therapeutics.
