Visual and automatic classification of the cyclic alternating pattern in electroencephalography during sleep.

Area of Science:

• Neurophysiology
• Sleep Medicine
• Biomedical Engineering

Background:

• Cyclic alternating pattern (CAP) is a recognized neurophysiological marker in sleep.
• Current visual scoring of CAP relies on international criteria and multiple EEG channels.
• Evaluating the efficiency and reliability of single-channel CAP scoring is essential for broader application.

Purpose of the Study:

• To determine the feasibility of visual CAP scoring using only one electroencephalogram (EEG) channel.
• To assess inter-scorer agreement for single-channel CAP scoring across diverse sleep recordings.
• To compare the agreement of visual scoring with automatic scoring systems.

Main Methods:

• Utilized 16 hours of single-channel sleep EEG recordings from four European sleep laboratories.
• Seven independent scorers performed visual CAP scoring based on international criteria.
• Two automatic blind scoring systems were employed for comparison.
• Event-based inter-scorer agreement analysis, including pairwise inter-scorer agreement (PWISA), was conducted.

Main Results:

• Pairwise inter-scorer agreement (PWISA) ranged from 55.5% to 84.3%, with an average PWISA above 60% for all scorers.
• The global average inter-scorer agreement was 69.9%.
• Automatic scoring systems demonstrated results comparable to visual scoring.

Conclusions:

• Cyclic alternating pattern (CAP) can be reliably scored using a single EEG channel.
• Single-channel CAP scoring exhibits sufficient inter-scorer agreement, supporting its clinical utility.
• The findings suggest that CAP scoring can be integrated into automated sleep analysis systems with performance similar to manual scoring.