Combining SNNs with Filtering for Efficient Neural Decoding in Implantable Brain-Machine Interfaces

TL;DR

This work tackles the data-rate bottleneck in wireless implantable brain-machine interfaces by advocating edge decoding and proposing a hybrid approach that combines traditional signal filtering with Spiking Neural Networks (SNNs). By applying block bidirectional Bessel filters to SNN outputs (and exploring Butterworth and Chebyshev alternatives), the authors achieve up to roughly $\\approx 8\\%$ improvements in $R^2$ for SNN topologies and significant gains for other models, narrowing the accuracy gap to LSTMs while maintaining low compute and memory demands. Across datasets and benchmarks (notably Neurobench), filtered SNNs offer state-of-the-art or near state-of-the-art performance with favorable resource efficiency, supporting the viability of decoder-integrated implants. The findings imply that combining edge filtering with neuromorphic decoding can deliver practical, energy-efficient iBMI solutions capable of handling tens to thousands of channels with reduced wireless data rates and preserved patient privacy.

Abstract

While it is important to make implantable brain-machine interfaces (iBMI) wireless to increase patient comfort and safety, the trend of increased channel count in recent neural probes poses a challenge due to the concomitant increase in the data rate. Extracting information from raw data at the source by using edge computing is a promising solution to this problem, with integrated intention decoders providing the best compression ratio. Recent benchmarking efforts have shown recurrent neural networks to be the best solution. Spiking Neural Networks (SNN) emerge as a promising solution for resource efficient neural decoding while Long Short Term Memory (LSTM) networks achieve the best accuracy. In this work, we show that combining traditional signal processing techniques, namely signal filtering, with SNNs improve their decoding performance significantly for regression tasks, closing the gap with LSTMs, at little added cost. Results with different filters are shown with Bessel filters providing best performance. Two block-bidirectional Bessel filters have been used--one for low latency and another for high accuracy. Adding the high accuracy variant of the Bessel filters to the output of ANN, SNN and variants provided statistically significant benefits with maximum gains of $\approx 5\%$ and $8\%$ in $R^2$ for two SNN topologies (SNN\_Streaming and SNN\_3D). Our work presents state of the art results for this dataset and paves the way for decoder-integrated-implants of the future.

Figures (24)

• Figure 1: (a) Conceptual figure of an iBMI that reads user intent and controls an effector. (b) Trend of exponential increase in the number of simultaneously recorded neuronsstevenson_2013. (c) Transmission data rates for the different cases (Raw data, spike sorting, decoding). While spike sorting provides some compression, decoding on the implant can provide the best option, especially as the number of electrodes increases beyond 1000. (d) Integrating computing in the implant can reduce the wireless datarate enabling scalability of iBMI systems.
• Figure 2: The experiment in the dataset has the NHP controlling the cursor and moving it to the target location. Once the NHP completes the action (referred to as a reach), the target location will move to a new location, and the subject will move the cursor accordingly.
• Figure 3: How each reach is defined in this work: a) The start and end of a reach are marked by the index where there is a change in the target location array, indicating the monkey has moved the cursor to the previous target location. b) A sample segment taken from the file indy_20160622_01, where we can see five consecutive reaches being segmented.
• Figure 4: Architecture of models used in this paper are a) ANN b) ANN_3D c) SNN_3D d) SNN_Streaming and e) LSTM.
• Figure 5: Input data pre-processing methods for feature extraction presented in this paper: a) Summation mode, where the number of spikes detected within a bin window $T_W$ for each probe is summed to create a feature. b) Sub-window mode, where the bin window is further divided into $m$ sub-windows, and the number of spikes detected within each sub-window is summed. c) Streaming mode, where the input spike is gathered as it is.