Large-scale brain recording is essential for understanding the dynamics of neural populations and for developing brain-machine interfaces (BMIs) for the rehabilitation of neurological disorders. However, a major limitation remains the difficulty of processing these massive data streams in real-time with low energy consumption. To address this, the thesis proposes implementing spiking neural networks (SNNs) on neuromorphic hardware to enable real-time neural signal processing. Built on a SoC FPGA, the developed platform enables real-time, low-latency analysis of the neural dynamics of cultured neurons, while remaining flexible, affordable, and accessible for integration into recording pipelines and bio-hybrid experiments.
