Data from: Eulerian videography technology improves classification of sleep architecture in primates
Melvin, Emilie M.; Samson, David R.; Nunn, Charles L. (2019), Data from: Eulerian videography technology improves classification of sleep architecture in primates, Dryad, Dataset, https://doi.org/10.5061/dryad.07sj8sq
Sleep is a critically important dimension of primate behavior, ecology, and evolution, yet primate sleep is under-studied because current methods of analyzing sleep are expensive, invasive, and time-consuming. In contrast to electroencephalography (EEG) and actigraphy, videography is a cost-effective and non-invasive method to study sleep architecture in animals. With video data, however, it is challenging to score subtle changes that occur in different sleep states, and technology has lagged behind innovations in EEG and actigraphy. Here, we applied Eulerian videography to magnify pixels relevant to scoring sleep from video, and then compared these results to analyses based on actigraphy and standard infrared videography. We studied four species of lemurs (Eulemur coronatus, Lemur catta, Propithecus coquereli, Varecia rubra) for 12-hour periods per night, resulting in 6,480 one-minute epochs for analysis. Cramer’s V correlation between actigraphy-classified sleep and infrared videography-classified sleep revealed consistent results in 8 out of 9 of the 12 h videos scored. A sample of the infrared videography was then processed by Eulerian videography for movement magnification and re-coded. A second Cramer’s V correlation analysis, between two independent scorers coding the same Eulerian-processed video, found that inter-observer agreement among Eulerian videography increased sleep vs. awake, NREM, and REM classifications by 7.1%, 46.7%, and 34.3%, respectively. Furthermore, Eulerian videography was more strongly correlated with actigraphy data when compared to results from standard infrared videography. The increase in agreement between the two scorers indicates that Eulerian videography has the potential to improve the identification of sleep states in lemurs and other primates, and thus to expand our understanding of sleep architecture without the need for EEG.