Validation of an automated sleep spindle detection method for mouse electroencephalography

Area of Science:

• Neuroscience
• Sleep Science
• Computational Biology

Background:

• Sleep spindles are crucial for cognitive function and are altered in neuropsychiatric disorders.
• Rodent models are valuable for studying the pathophysiology of spindle abnormalities.
• Reliable detection of sleep spindles in mouse electroencephalography (EEG) remains a challenge.

Purpose of the Study:

• To develop and validate an automated algorithm for accurate sleep spindle detection in mouse EEG.
• To overcome the unreliability of manual spindle scoring in raw mouse EEG data.
• To provide a robust tool for investigating spindle generation mechanisms and alterations in disease models.

Main Methods:

• Manual spindle detection procedures were optimized to create an automated algorithm for mouse cortical EEG.
• Algorithm accuracy was validated against sigma band power analysis and pharmacological manipulations.
• Manual scoring of processed EEG data (filtered, cubed root-mean-squared) was compared to raw EEG scoring and the algorithm.

Main Results:

• Manual identification of sleep spindles in raw mouse EEG was unreliable (F1-score 0.26 ± 0.07).
• Reliable manual scoring was achieved with processed EEG data (F1-score > 0.95), matching algorithm performance.
• Algorithmically detected spindles correlated with sigma-power and were sensitive to sleep-wake states, sleep stage transitions, and zolpidem administration.

Conclusions:

• An automated, validated paradigm for rapid and reliable detection of sleep spindles from mouse EEG recordings has been established.
• This technique offers a powerful tool for studying spindle generation mechanisms.
• It facilitates research into spindle alterations observed in mouse models of neuropsychiatric disorders.