Noninvasive Sleep Scoring in Mice using Electric Field Sensors

Abstract

29Background: Rodent sleep scoring in principally reliant on electroencephalogram (EEG) and 30 electromyogram (EMG), but this approach is invasive, can be expensive, and requires expertise 31 and specialized equipment. Affordable, simple to use, and noninvasive ways to accurately 32 quantify rodent sleep are needed. 34New method: We developed and validated a new method for sleep-wake staging in mice using 35 cost-effective, noninvasive electric field (EF) sensors that detect respiration and other 36 movements. We validated recordings from EF sensors attached to the exterior of specialty 37 chambers used to continuously capture sleep with EEG/EMG, then compared this to EF sensors 38 attached to vivarium home-cages. 39 40 Results: EF sensors quantified 3-state sleep architecture (wake, rapid eye movement -REM -41 sleep, and non-REM sleep) with high agreement (>93%) and comparable inter-and intra-scorer 42 error as expert EEG/EMG scoring. Novices given an instruction document with examples were 43 able to score sleep comparable to expert scorers (>91% agreement). Additionally, EF sensors 44 were able to quantify 3-state sleep scoring in traditional mouse home cages. 45 46 Comparison with existing method: Most noninvasive sleep assessment technology requires 47 animal contact, altered cage environments, and/or can only discern 2 states of arousal (wake or 48 asleep). The EF sensors are able to discriminate REM from non-REM sleep accurately and from 49 outside the animal's home cage.50 51 Conclusions: EF sensors provide a simple and reliable method to accurately score 3-state sleep 52 architecture; (i) from outside the typical home cage, (ii) where noninvasive approaches are 53 preferred, or (iii) which EEG/EMG is not possible.54 3 55 56 Graphical Abstract: 57 58 59 60 Accurately characterizing sleep-wake is crucial to understanding its impact on health, cognition, 61 and injury recovery. 1-3 Sleep staging in humans, relies on substantial non-invasive 62 instrumentation (polysomnography) including electroencephalogram (EEG), electromyogram 63 (EMG) of chin and limbs, electrooculogram (EOG) to detect rapid and rolling eye movements, 64 along with chest wall or air flow movement, oximetry, and video. 4-6 However, rodent studies 65 typically rely on fewer signals, invasive implanted EEG and EMG, less commonly invasive EOG 66 electrodes, and video analysis. Invasive surgical implants can result in weight loss, extensive 67 recovery, and commonly requires use of a tether cable. 7-9 Moreover, surgical expertise, limited 68 locations for electrode placement, and specialized equipment required to collect EEG and EMG 69 data may restrict inclusion of sleep analysis in an experimental design. For these reasons, there 70 is a need to develop noninvasive methods to assess sleep in preclinical rodent models. 71 72 Most noninvasive rodent sleep assessment methods center around measures of gross body 73 movement. Several video analysis techniques have been used to distinguish between 2-state 74 conditions (wake or asleep), but are u...

Classifier for the Rapid Simultaneous Determination of Sleep-Wake States and Seizures in Mice

“…As a benchmark, we choose the classification accuracy seen with human scoring as determined by inter-rater agreement. The agreement between scorers in our laboratory is high (>93%) (Kloefkorn et al, 2020), in accord with other reports of 92% (Rytkönen et al, 2011). Here we describe a highly accurate method for simultaneous automated sleep-wake and seizure classification.…”

“…These time savings are also not traded off for any loss of precision in the sleep scores. The scoring accuracies of the classifier, with a weighted average of > 96%, exceed the 93% agreement in human scores (Kloefkorn et al, 2020) for AASM/Rechtshaffen and Kales sleep scoring, making it a well-rounded classifier for sleep.…”

“…These large datasets often involve thousands of hours of multi-channel recording. Manual scoring of this data for both seizures and sleep-wake is often impractical; a trained expert scorer may take up to 25 minutes to score sleep-wake for 12 hours of data (Kloefkorn et al, 2020), and even longer when EEGs are abnormal. Conventional algorithmic and parametric sleep-wake scoring methods are not applicable as these approaches typically depend on the ratio of EEG power in the theta and delta bands (Costa-Miserachs et al, 2003), and perform poorly with variably abnormal EEGs.…”

Relating spinal injury-induced neuropathic pain and spontaneous afferent activity to sleep and respiratory dysfunction

“…Sleep was analyzed from the EF sensor home cage recordings. As described previously 44,46 , wake, rapid eye movement (REM) sleep, and non-REM sleep were quantified through changes in animal movement and respiration (Figure 1B). Weekly averages were calculated for total sleep time (TST), percent time spent in each arousal state, average sleep bout duration (SBD), average REM sleep event duration, REM sleep latency, percentage of sleep bouts containing REM sleep, sleep fragmentation index (SFI: total awakenings per hour of sleep), and micro arousal index (MAI: total micro arousals < 60 seconds per hour of sleep).…”