An Unsorted Spike-Based Pattern Recognition Method for Real-Time Continuous Sensory Event Detection from Dorsal Root...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Functional neuromuscular stimulation (FNS) systems aim to restore function in patients with paralysis.
• Closed-loop control using sensory feedback is crucial for improving FNS system efficacy.
• Current methods for real-time sensory event detection from afferent signals are limited.

Purpose of the Study:

• To develop and evaluate a novel unsorted spike-based pattern recognition method for continuous, real-time detection of sensory events from tactile afferent signals.
• To assess the feasibility of this method for sensory feedback in closed-loop FNS systems.

Main Methods:

• Recording of tactile afferent signals using a 16-channel microelectrode in the dorsal root ganglion.
• Extraction of unsorted spike-based feature vectors combining time and time-frequency domain features.
• Dimensionality reduction using Principal Component Analysis (PCA).
• Classification of sensory events using a Multilayer Perceptron (MLP) classifier.

Main Results:

• The proposed method demonstrated good classification accuracy for sensory event detection.
• The processing time delay for sensory event detection was successfully maintained below 200 milliseconds.
• Feature vectors were effectively reduced in dimensionality using PCA.

Conclusions:

• The novel unsorted spike-based pattern recognition method is effective for real-time sensory event detection.
• The low processing time delay makes the method suitable for sensory feedback in closed-loop FNS systems.
• This approach holds potential for improving functional recovery in patients with hemiplegia or quadriplegia.