Deciphering the Morphology of Motor Evoked Potentials.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Clinical Neurology

Background:

• Motor Evoked Potentials (MEPs) are crucial for monitoring disability progression in Multiple Sclerosis (MS).
• Current definitions of normal vs. abnormal MEP morphology lack standardization, hindering consistent clinical application.
• Expert interpretation of MEP morphology shows agreement on the concept but variability in classification thresholds.

Purpose of the Study:

• To establish an objective and reproducible method for defining MEP morphology.
• To identify a reliable automated feature that serves as a proxy for MEP morphological abnormality.
• To validate this automated feature against expert consensus for clinical utility in MS.

Main Methods:

• Five experts independently labeled 1,000 MEPs for morphology (normal/abnormal).
• 5,943 time series features were extracted from MEPs.
• One-dimensional logistic regression models were used to assess feature performance in reproducing expert labels, with Approximate Entropy (ApEn) identified as a key feature.

Main Results:

• Approximate Entropy (ApEn) accurately reproduced the majority-vote labels assigned by neurologists.
• ApEn achieved an AUC score of 0.92 on an independent test set, outperforming the average neurologist in reproducing consensus.
• The study demonstrates that pooling expert interpretations provides a consistent definition of MEP morphology.

Conclusions:

• MEP morphology can be consistently defined by aggregating multiple expert interpretations.
• Approximate Entropy (ApEn) is a valid, continuous score for quantifying MEP morphological abnormality.
• An automated, objective MEP morphology score facilitates large-scale research, multi-center studies, and clinical adoption for MS patient follow-up.