Spiking Neural Network for Intra-cortical Brain Signal Decoding

2025-04-12

AI Summary

Traditional vector-based methods for intracortical brain–computer interfaces (BCIs) suffer from limited decoding accuracy, while deep learning models exhibit poor energy efficiency—posing critical bottlenecks for real-time, implantable systems. To address this, we propose the first spiking neural network (SNN) framework specifically designed for low-latency, real-time neural decoding. Our approach introduces a novel cross-modal fusion mechanism that integrates handcrafted neural activity vectors with learned deep features, synergistically leveraging event-driven computation and lightweight feature engineering to achieve high accuracy without compromising energy efficiency. Evaluated on motor decoding tasks in two rhesus monkeys, our SNN outperforms conventional artificial neural networks (ANNs) in accuracy while delivering 10× to 100× improvements in inference energy efficiency. This work marks the first successful deployment of SNNs for real-time intracortical BCI decoding, establishing a new paradigm for high-accuracy, ultra-low-power implanted BCIs.

Technology Category

Cognitive Modeling & Cognitive Systems: Neural Spike CodingMachine Learning: Multimodal LearningHumans and AI: Brain-Sensing and Analysis

Application Category

Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphsSearch and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingSystems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Decoding brain signals accurately and efficiently is crucial for intra-cortical brain-computer interfaces. Traditional decoding approaches based on neural activity vector features suffer from low accuracy, whereas deep learning based approaches have high computational cost. To improve both the decoding accuracy and efficiency, this paper proposes a spiking neural network (SNN) for effective and energy-efficient intra-cortical brain signal decoding. We also propose a feature fusion approach, which integrates the manually extracted neural activity vector features with those extracted by a deep neural network, to further improve the decoding accuracy. Experiments in decoding motor-related intra-cortical brain signals of two rhesus macaques demonstrated that our SNN model achieved higher accuracy than traditional artificial neural networks; more importantly, it was tens or hundreds of times more efficient. The SNN model is very suitable for high precision and low power applications like intra-cortical brain-computer interfaces.

Problem

• Improving accuracy and efficiency in intra-cortical brain signal decoding
• Reducing computational cost of deep learning-based brain signal decoding
• Enhancing decoding performance for brain-computer interface applications

Innovation

• Spiking Neural Network for efficient decoding
• Feature fusion enhances decoding accuracy
• Energy-efficient model for brain-computer interfaces