Related Experiment Videos
Real-time automated sleep scoring: validation of a microcomputer-based system for mice
R N Van Gelder1, D M Edgar, W C Dement
1Sleep Disorders and Research Center, Stanford University School of Medicine, California 94304.
Sleep
|February 1, 1991
Summary
A new computer system, SCORE, automates rodent sleep scoring from polygraph records. This system significantly reduces labor for long-term circadian studies, achieving high accuracy in distinguishing sleep and wakefulness states.
Area of Science:
- Neuroscience
- Chronobiology
- Computational Biology
Background:
- Long-term rodent sleep studies are labor-intensive due to manual analysis of polygraph records.
- Automated analysis is needed to expedite circadian research.
Purpose of the Study:
- To develop and validate SCORE, a microcomputer-based system for real-time rodent sleep scoring.
- To reduce the manual labor involved in analyzing long-term sleep and wakefulness data.
Main Methods:
- Digitizing electroencephalograph (EEG) data in 10-s epochs at 100 Hz.
- Extracting waveform features into a 48-dimension vector for comparison with canonical arousal states (wakefulness, theta-dominated wakefulness, REM sleep, NREM sleep).
- Scoring epochs based on the best match value after excluding activity data (wheel-running, drinking).
Main Results:
- SCORE achieved 94.0% agreement with human scorers across over 23,000 epochs in four mice.
- High accuracy was observed for wakefulness (97.8%) and NREM sleep (94.7%).
- Lower accuracy was noted for REM sleep (75.2%) and theta-dominated wakefulness (83.3%), with REM errors near actual REM epochs.
Conclusions:
- SCORE effectively automates rodent sleep scoring, significantly facilitating long-term circadian studies.
- The system can score arousal states for eight animals simultaneously in real time.
- SCORE offers a valuable tool for researchers studying sleep and wakefulness patterns in rodents.