Principal Component Regression on Motor Evoked Potential in Single-Pulse Transcranial Magnetic Stimulation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Motor Control

Background:

• Motor evoked potentials (MEPs) are crucial for assessing corticospinal excitability.
• Traditional MEP analysis often overlooks detailed morphological information, focusing only on latency and amplitude.
• This limitation hinders a comprehensive understanding of neural pathway function and variability.

Purpose of the Study:

• To investigate the utility of principal component regression (PCR) for analyzing MEP morphology.
• To explore PCR's potential for enhancing automated quantification and data reduction in MEP analysis.
• To determine the minimum number of trials needed for reliable MEP dataset representation.

Main Methods:

• MEPs were recorded from the first dorsal interosseous muscle using navigated transcranial magnetic stimulation (TMS) targeting the primary motor cortex.
• Principal component regression (PCR) was applied to parameterize MEPs and capture at least 96% of the total variance.
• Non-linear least square estimation was used to reconstruct original MEPs from principal components (PCs).

Main Results:

• PCR effectively characterized MEP morphology, summarizing redundant information into a principal component set.
• The method demonstrated robustness by removing environmental noise and neuronal pathway inconsistencies, aiding automated quantification.
• Analysis revealed that approximately 20 MEP trials capture about 90% of the dataset's total variance, indicating data saturation.

Conclusions:

• Principal component regression offers a powerful approach to extract comprehensive morphological information from MEPs.
• PCR improves the reliability and efficiency of automated MEP analysis and noise reduction.
• The findings provide a data-driven guideline for optimizing trial numbers in MEP studies for robust results.