Spike sorting paradigm for classification of multi-channel recorded fasciculation potentials

Area of Science:

• Neuroscience
• Computational Biology
• Biomedical Engineering

Background:

• Fasciculation potentials (FPs) are crucial for diagnosing Amyotrophic Lateral Sclerosis (ALS).
• FP shape classification offers valuable insights for motor unit neurophysiology.
• Current methods may lack efficiency in handling the irregular nature of FP firings.

Purpose of the Study:

• To develop and evaluate a Matlab program for classifying fasciculation potentials (FPs).
• To enhance the electrodiagnostic process for Amyotrophic Lateral Sclerosis (ALS) through improved FP analysis.
• To provide an efficient tool for laboratory-based neurophysiological investigations.

Main Methods:

• Utilized multi-channel surface electromyogram (EMG) electrodes for FP recording.
• Applied Principal Component Analysis (PCA) for feature extraction across channels.
• Implemented unsupervised and supervised classification algorithms for FP sorting.
• Incorporated interactive, manual modification for progressive accuracy improvement.

Main Results:

• Tested on 10 ALS patient datasets using a 20-channel electrode array.
• Successfully detected and classified 11,891 FPs into 235 distinct template waveforms.
• Demonstrated efficient classification and evaluation, with large datasets manageable within 1-2 days.
• Interactive features significantly expedited the refinement of classification outcomes.

Conclusions:

• The developed Matlab program serves as an efficient and effective toolbox for FP classification.
• The program facilitates accurate and timely analysis of FPs in neurophysiological studies.
• This tool has the potential to improve the electrodiagnosis of neurological disorders like ALS.