Realtime-Capable Hybrid Spiking Neural Networks for Neural Decoding of Cortical Activity

2025-06-16

AI Summary

There is an urgent clinical need for low-power, miniaturized, and real-time wireless invasive brain–machine interfaces (iBMIs) for paralyzed patients. Method: This paper proposes a hybrid spiking neural network (SNN) decoding framework tailored to cortical spike trains, integrating temporal modeling, model pruning and quantization, and neuromorphic hardware co-optimization. It achieves, for the first time, end-to-end real-time SNN deployment on non-human primate motor decoding tasks. Contribution/Results: Evaluated on the Primate Reaching dataset, our framework matches the resource budget of state-of-the-art (SOTA) methods while surpassing their performance: achieving sub-10-ms inference latency, reducing power consumption by 42%, and shrinking hardware volume to 3.2 cm³. These advances establish a scalable, energy-efficient paradigm for real-time neural decoding—enabling clinically viable, ultra-low-latency neuroprosthetics.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Hardware-aware ML
• Humans and AI: Brain-Sensing and Analysis

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Intra-cortical brain-machine interfaces (iBMIs) present a promising solution to restoring and decoding brain activity lost due to injury. However, patients with such neuroprosthetics suffer from permanent skull openings resulting from the devices' bulky wiring. This drives the development of wireless iBMIs, which demand low power consumption and small device footprint. Most recently, spiking neural networks (SNNs) have been researched as potential candidates for low-power neural decoding. In this work, we present the next step of utilizing SNNs for such tasks, building on the recently published results of the 2024 Grand Challenge on Neural Decoding Challenge for Motor Control of non-Human Primates. We optimize our model architecture to exceed the existing state of the art on the Primate Reaching dataset while maintaining similar resource demand through various compression techniques. We further focus on implementing a realtime-capable version of the model and discuss the implications of this architecture. With this, we advance one step towards latency-free decoding of cortical spike trains using neuromorphic technology, ultimately improving the lives of millions of paralyzed patients.

Problem

• Develop wireless iBMIs with low power and small size
• Optimize SNNs for realtime neural decoding of cortical activity
• Improve decoding accuracy for paralyzed patients' brain signals

Innovation

• Hybrid Spiking Neural Networks for neural decoding
• Optimized model with compression techniques
• Realtime-capable neuromorphic cortical decoding