Efficient Detection of Cortical Auditory Evoked Potentials in Adults Using Bootstrapped Methods.

Area of Science:

• Auditory Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Statistical detection methods aid in cortical auditory evoked potential (CAEP) analysis but often rely on assumptions about electroencephalogram (EEG) data.
• Violated assumptions can lead to reduced sensitivity and inaccurate false-positive rates in CAEP detection.
• Bootstrap methods offer an alternative, reducing reliance on distributional assumptions and allowing greater flexibility in statistical test design.

Purpose of the Study:

• To enhance the reliability and efficiency of CAEP detection methods.
• To improve the specificity and sensitivity of objective CAEP detection.

Main Methods:

• Evaluated Hotelling's T2 test, Fmp, modified q-sample statistics, and template-based methods (correlation, covariance, dynamic time-warping - DTW).
• Utilized both simulations and CAEP threshold data from 23 normal-hearing adults.
• Employed bootstrapping for significance evaluation, particularly with DTW.

Main Results:

• Dynamic time-warping (DTW) evaluated using bootstrap demonstrated the highest sensitivity.
• DTW achieved up to a 30% increase in test sensitivity compared to the conventional Hotelling's T2 test.
• Effective DTW performance depends on the template correlating well with the subject's CAEP morphology.

Conclusions:

• Dynamic time-warping (DTW) is a highly sensitive method for CAEP detection when subject CAEP morphology is known beforehand.
• The bootstrap-evaluated DTW method significantly improves upon traditional statistical approaches for CAEP analysis.
• This approach enhances objective CAEP detection, benefiting clinical applications.