Spatial Information Based OSort for Real-Time Spike Sorting Using FPGA

Objective: Spiking activity of individual neurons can be separated from the acquired multi-unit with spike sorting methods. Processing recorded high-dimensional neural data take a large amount time when performed on general-purpose computers. Methods: In this paper, an FPGA-based real-time system is presented which takes into account spatial correlation between electrical signals closely-packed recording sites to cluster multi-channel data. The uses window-based version Online Sorting algorithm, unsupervised template-matching for clustering. Results: test results show that proposed reach average accuracy 86% using simulated (16-32 neurons, 4-10 dB Signal-to-Noise Ratio), while single-channel clustering achieves only 74% in same cases 128-channel electrode array. developed was also tested vivo cortical recordings obtained anesthetized rat. Conclusion: process more than 11000 spikes/second, so it used during experiments providing feedback location and electrophysiological properties well-separable single units. Significance: could reduce positioning error site measurement.