Real-Time FPGA-Based Multichannel Spike Sorting Using Hebbian Eigenfilters

Abstract

Abstract-Real-time multi-channel neuronal signal recording has spawned broad applications in neuro-prostheses and neurorehabilitation. Detecting and discriminating neuronal spikes from multiple spike trains in real-time require significant computational efforts and present major challenges for hardware design in terms of hardware area and power consumption. This paper presents a Hebbian eigenfilter spike sorting algorithm, in which principal components analysis (PCA) is conducted through Hebbian learning. The eigenfilter eliminates the need of computationally expensive covariance analysis and eigenvalue decomposition in traditional PCA algorithms and, most importantly, is amenable to low cost hardware implementation. Scalable and efficient hardware architectures for real-time multi-channel spike sorting are also presented. In addition, folding techniques for hardware sharing are proposed for better utilization of computing resources among multiple channels. The throughput, accuracy and power consumption of our Hebbian eigenfilter are thoroughly evaluated through synthetic and real spike trains. The proposed Hebbian eigenfilter technique enables real-time multi-channel spike sorting, and leads the way towards the next generation of motor and cognitive neuro-prosthetic devices.Index Terms-Brain-machine interface, Hebbian learning, spike sorting, FPGAs, hardware architecture design.

Complexity Optimization and High-Throughput Low-Latency Hardware Implementation of a Multi-Electrode Spike-Sorting Algorithm

“…Most of the presently available devices that feature real-time performance are based on simple decision rules of different threshold crossings [12], [13], which provide a low-complexity and low-latency implementation, but cannot, however, guarantee a high spike-sorting performance. One of the spike-sorting methods, commonly used as a benchmark method, principal component analysis (PCA), has been optimized for real-time implementation in the devices presented in [14]–[16]. However, all of these methods target single-electrode-based devices (i.e., devices in which single neurons are sensed by single electrodes).…”

Stream-based Hebbian eigenfilter for real-time neuronal spike discrimination

“…It has been shown in the experiments that the convergence of the eigenfilter is relatively independent to the initial synaptic weights. However, the learning rates would contribute to the speed of the convergence, as reported in [ 24 ]. Both true positive rate and false positive rate of classification were considered in our evaluation, as shown in Figure 6(a) and 6(b) .…”

“…In our previous work [24], general Hebbian algorithm (GHA), so-called Hebbian eigenfilter, which presents an efficient approach for realizing PCA, was proposed for calculating the leading PCs of neuronal spikes. Let $\overrightarrow{text\mathsf{\text{x}}}text\mathsf{\text{(i)}}$, i ∈ [1, n] be n aligned spikes.…”

Neural Spike Sorting Using Binarized Neural Networks

“…The learning process of the employed BNN is performed offline, which entails a partially supervised spike sorting. Most neural signal processing systems involve some offline processing [2], [3], [5], [9], [18], [19], which results in a more area-and This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ power-efficient realization.…”