Explainable Spike Detection Algorithm on C-fiber Microneurography Data using Weighted Hermite Variable Projection...

Area of Science:

• Neuroscience
• Artificial Intelligence
• Signal Processing

Background:

• Microneurography is crucial for understanding peripheral nerve function and sensory perception.
• Detecting spike events in microneurography data is challenging due to low signal-to-noise ratios and class imbalance.
• Accurate spike detection is vital for interpreting neurophysiological patterns and understanding mechanisms of pain and itch.

Purpose of the Study:

• To develop and validate a novel neural network-based approach for automated spike detection in microneurography data.
• To address challenges of low signal-to-noise ratio and class imbalance in microneurography signals.
• To provide an interpretable and accurate solution for analyzing peripheral nerve activity.

Main Methods:

• Utilized a weighted Hermite Variable Projection Neural Network model for spike event detection.
• Implemented a knowledge-augmented model, data balancing techniques, and explainable AI methods.
• Applied a sliding window technique for time series signal analysis and continuous detection.

Main Results:

• The proposed pipeline successfully detects spike events in microneurography data with enhanced accuracy and interpretability.
• The explainable AI methods revealed the neural patterns and feature extraction mechanisms employed by the model.
• The system provides a fully automated solution for spike detection, overcoming previous limitations.

Conclusions:

• This work presents the first neural network-based solution for microneurography spike detection.
• The developed method offers a reliable and interpretable tool for analyzing peripheral neural activity.
• This automated approach has the potential to improve data interpretation and advance research in pain and itch mechanisms.