Automatic sleep-wake scoring in the rat on microcomputer APPLE II

Sleep scoring in rodents: Criteria, automatic approaches and outstanding issues

There is nothing we spend as much time on in our lives as we do sleeping, which makes it even more surprising that we currently do not know why we need to sleep. Most of the research addressing this question is performed in rodents to allow for invasive, mechanistic approaches. However, in contrast to human sleep, we currently do not have shared and agreed upon standards on sleep states in rodents. In this article, we present an overview on sleep stages in humans and rodents and a historical perspective on the development of automatic sleep scoring systems in rodents. Further, we highlight specific issues in rodent sleep that also call into question some of the standards used in human sleep research.

The pros and cons of using automated sleep scoring in sleep research

Sleep scoring plays a pivotal role both in sleep research and in clinical practice. Traditionally, this process has relied on manual scoring by human experts, but it is marred by time constraints, and inconsistencies between different scorers. Consequently, the quest for more efficient and reliable approaches has sparked a great interest in the realm of automatic sleep-scoring methods. In this article, we provide an exploration of the merits and drawbacks of automatic sleep scoring, alongside the pressing challenges and critical considerations that demand attention in this evolving field.

Identification of the sleep-wake states in rats using the high-frequency activity of the electroencephalogram

The electroencephalographic signal constitutes the main sign classically used for the identification of states of alertness. However, activities in the high frequency (>100 Hz) range have not been properly studied despite their high potential for sleep scoring in rodents. In the present study, we designed a method for the identification of the sleep-wake states in rats by exclusively using high-frequency activities of the electroencephalogram. By calculating the ratio between the amplitude of the electroencephalographic signal from 110 to 200 Hz and from 110 to 300 Hz, we obtained an index that had values that were low during wakefulness, intermediate during non-REM sleep and high during REM sleep. This high-frequency index (HiFI) allowed the identification of each state without the need to study other signs such as muscle activity or eye movements. To evaluate the performance of the index, we compared it with the conventional scoring of the sleep-wake cycle based upon the study of the electromyogram and delta (0.5-4 Hz), theta (6-9 Hz) and sigma (10-14 Hz) bands of the electroencephalogram. The index had an accuracy of 90.43 ± 1.91% (Cohen's kappa value of 0.82), confirming that the study of the high-frequency activities of the electroencephalogram was sufficient to reliably identify alertness states in the rat. Compared to other sleep-scoring methods, the HiFI has several advantages. It only requires one electroencephalography electrode, thus reducing the severity of the surgical preparation of the experimental animal, and its calculation is very simple, so it can be easily implemented online to classify sleep-wake states in real time.

Acute levodopa dosing around-the-clock ameliorates REM sleep without atonia in hemiparkinsonian rats

Rapid-eye-movement (REM) sleep without atonia (RSWA), a marker of REM sleep behavior disorder (RBD), is frequently comorbid with Parkinson's disease (PD). Although rodent models are commonly used for studying PD, the neurobiological and behavioral correlates of RBD remain poorly understood. Therefore, we developed a behavior-based criteria to identify RSWA in the hemiparkinsonian rat model of PD. Video recordings of rats were analyzed, to develop a criteria consisting of behavioral signs that occurred during polysomnographically confirmed epochs of sleep-wake stages. The sleep-slouch, a postural shift of the body or head caused only by gravity, was identified as a unique behavioral sign of REM sleep onset and was altered in hemiparkinsonian rats during RSWA. There was a significant correlation between the behavior-based criteria and polysomnograms for all sleep-wake stages in control but not hemiparkinsonian rats indicating a deterioration of sleep-wake architecture in parkinsonism. We then tested the efficacy of levodopa in ameliorating RSWA using intermittent and around-the-clock (ATC) dosing regimens. ATC levodopa dosing at 4 mg/kg for 48 h caused a significant reduction of RSWA as measured by polysomnography and the behavioral-based criteria along with an amelioration of forelimb motor deficits. Our findings show that the phenomenological correlates of RSWA can be reliably characterized in the hemiparkinsonian rat model. ATC levodopa administration ameliorates RSWA in this model without deleterious consequences to the overall sleep-wake architecture and therapeutic benefits for parkinsonian motor deficits. These findings suggest that further study may allow for the application of a similar approach to treat RBD in PD patients.

Unsupervised online classifier in sleep scoring for sleep deprivation studies

STUDY OBJECTIVE

This study was designed to evaluate an unsupervised adaptive algorithm for real-time detection of sleep and wake states in rodents.

DESIGN

We designed a Bayesian classifier that automatically extracts electroencephalogram (EEG) and electromyogram (EMG) features and categorizes non-overlapping 5-s epochs into one of the three major sleep and wake states without any human supervision. This sleep-scoring algorithm is coupled online with a new device to perform selective paradoxical sleep deprivation (PSD).

SETTINGS

Controlled laboratory settings for chronic polygraphic sleep recordings and selective PSD.

PARTICIPANTS

Ten adult Sprague-Dawley rats instrumented for chronic polysomnographic recordings.

MEASUREMENTS

The performance of the algorithm is evaluated by comparison with the score obtained by a human expert reader. Online detection of PS is then validated with a PSD protocol with duration of 72 hours.

RESULTS

Our algorithm gave a high concordance with human scoring with an average κ coefficient > 70%. Notably, the specificity to detect PS reached 92%. Selective PSD using real-time detection of PS strongly reduced PS amounts, leaving only brief PS bouts necessary for the detection of PS in EEG and EMG signals (4.7 ± 0.7% over 72 h, versus 8.9 ± 0.5% in baseline), and was followed by a significant PS rebound (23.3 ± 3.3% over 150 minutes).

CONCLUSIONS

Our fully unsupervised data-driven algorithm overcomes some limitations of the other automated methods such as the selection of representative descriptors or threshold settings. When used online and coupled with our sleep deprivation device, it represents a better option for selective PSD than other methods like the tedious gentle handling or the platform method.

Design and validation of a computer-based sleep-scoring algorithm

A computer-based sleep scoring algorithm was devised for the real time scoring of sleep-wake state in Wistar rats. Electroencephalogram (EEG) amplitude (microV(rms)) was measured in the following frequency bands: delta (delta; 1.5-6 Hz), theta (Theta; 6-10 Hz), alpha (alpha; 10.5-15 Hz), beta (beta; 22-30 Hz), and gamma (gamma; 35-45 Hz). Electromyographic (EMG) signals (microV(rms)) were recorded from the levator auris longus (neck) muscle, as this yielded a significantly higher algorithm accuracy than the spinodeltoid (shoulder) or temporalis (head) muscle EMGs (ANOVA; P=0.009). Data were obtained using either tethers (n=10) or telemetry (n=4). We developed a simple three-step algorithm that categorizes behavioural state as wake, non-rapid eye movement (NREM) sleep, rapid eye movement (REM) sleep, based on thresholds set during a manually-scored 90-min preliminary recording. Behavioural state was assigned in 5-s epochs. EMG amplitude and ratios of EEG frequency band amplitudes were measured, and compared with empirical thresholds in each animal.STEP 1: EMG amplitude greater than threshold? Yes: "active" wake, no: sleep or "quiet" wake. STEP 2: EEG amplitude ratio (delta x alpha)/(beta x gamma) greater than threshold? Yes: NREM, no: REM or "quiet" wake. STEP 3: EEG amplitude ratio Theta(2)/(delta x alpha) greater than threshold? Yes: REM, no: "quiet" wake. The algorithm was validated with one, two and three steps. The overall accuracy in discriminating wake and sleep (NREM and REM combined) using step one alone was found to be 90.1%. Overall accuracy using the first two steps was found to be 87.5% in scoring wake, NREM and REM sleep. When all three steps were used, overall accuracy in scoring wake, NREM and REM sleep was determined to be 87.9%. All accuracies were derived from comparisons with unequivocally-scored epochs from four 90-min recordings as defined by an experienced human rater. The algorithms were as reliable as the agreement between three human scorers (88%).

Automated sleep staging in rat with a standard spreadsheet

A new method of automated sleep-wake staging in the rat is described. Hippocampal electroencephalographic (HPC) and nuchal electromyographic signals were recorded by a digital polygraph. The HPC channel was filtered off-line to obtain the original plus theta and delta waves. Statistics of each of these four channels were obtained every 5 s and exported to a standard spreadsheet. The automated staging consisted of five steps: (1) automatic detection of waking, nonrapid eye movement sleep and rapid eye movement sleep patterns (5-s periods); (2) calculation of statistics for each vigilance state; (3) final classification of 5-s periods; (4) construction of a primary 20-s epoch hypnogram; and (5) automatic refinement of the previous hypnogram. The system includes indices about the accuracy of the staging and was validated with five recordings of 23 h each. The global agreement between human and automatic scoring in the validation recordings was 94.32%.

Automated sleep staging systems in rats

One of the major inconveniences encountered in sleep studies is the time consuming labor involved in equating visual analysis of physiological recordings (EEG, EMG, EOG, ...) to an appropriate state of vigilance. The explosion of computer technology is responsible for the emergence of several automated sleep-wake staging systems to supplement human analysis. Conversely to human sleep analysis, rat sleep is characterized by the absence of consensus about numerous elements constituting the sleep-wake staging systems used to build a hypnogram (recording position, length of epoch, number and definition of the vigilance state discriminated, ...). If justified, the choices of the parameters involved by each system generally result from various viewpoints (physiology, mathematics, electronics, ...). The diversity generated by the liberty offered the investigator in building a system excludes any rigorous comparison between systems. Nevertheless, this variety can also be viewed as a representative of the effervescence of research in the field of sleep, and as a catalyst for new ideas.

Influence of a GABA(B) receptor antagonist on the sleep-waking cycle in the rat

The influence of CGP 35348 (a GABA(B) receptor antagonist) on the sleep-waking cycle was studied in rats. The animals were injected i.p. at the beginning of the light period and the data expressed by 2-h periods and total duration (6 h). At 100 mg/kg, slow-wave sleep (SWS) was decreased during the 6-h recording with a peculiar decrease during the first 2 h. SWS was subdivided into three stages: slow-waves; spindles occurring as SWS deepens; and intermediate stage appearing prior to paradoxical sleep (PS). Only the slow-wave stage and intermediate stage were decreased. Waking was increased during the 6-h recording. It was subdivided into waking with hippocampal theta rhythm (psychomotor active waking) and waking without theta activity (quiet waking). Both were increased during the first 2 h. However, quiet waking was increased throughout the recording duration. At 300 mg/kg, SWS was decreased during the three 2-h periods. This decrease was principally related to a decrease of the slow-wave stage. PS was increased over the 6-h recording with a marked increase during the second 2-h period. Consequently, under the influence of the GABA(B) receptor antagonist, the SWS was decreased at the expense of behavioral stages with cortical low-voltage activity (waking and PS). GABAergic neurons are present in the mesopontine structures responsible for these two stages. We can conclude that endogenous GABA acting at the GABA(B) receptor level participates in the regulation of waking and PS.

Sleep cycle disturbances induced by apamin, a selective blocker of Ca(2+)-activated K+ channels

Intracerebroventricular injections of low doses of apamin, a specific blocker of a class of Ca(2+)-activated K+ channels, induced insomnia and a long-lasting suppression of deep slow sleep and paradoxical sleep. Injected animals showed a late but important rebound of paradoxical sleep. Even after the recovery of a normal sleep amount, the circadian cycle remained disturbed during all the recording duration (96 h).

Adult rat vigilance states discrimination by artificial neural networks using a single EEG channel

Two multilayer neural networks were designed to discriminate vigilance states (waking, paradoxical sleep, and non-REM sleep) in the rat using a single parieto-occipital EEG derivation. After filtering (bandwidth 3.18-25 Hz) and digitization at 512 HZ, the EEG signal was segmented into eight second epochs. Five variables (three statistical, two temporal) were extracted from each epoch. The first network computed an epoch by epoch classification, while the second network also utilized contextual information from contiguous epochs. A specific postprocessing procedure was developed to enhance the vigilance state discrimination of the neural networks designed and especially paradoxical sleep state estimation. The classifications made by the networks (with or without the postprocessing procedure) for six rats were compared to these made by two human experts using EMG and EEG informations on 63,000 epochs. High rates of agreement (> 90%) between humans and neural networks classifications were obtained. In view of its development possibilities and its applicability to other signals, this method could prove of value in biomedical research.

The transition from slow-wave sleep to paradoxical sleep: evolving facts and concepts of the neurophysiological processes underlying the intermediate stage of sleep

Paradoxical sleep in rats, cats and mice is usually preceded and sometimes followed by a short-lasting (a few seconds) electroencephalogram (EEG) stage characterized by high-amplitude spindles in the anterior cortex and low-frequency theta rhythm in the dorsal hippocampus. The former is an index of advanced slow-wave sleep; the latter is an index of limbic activation since it occurs during active waking and paradoxical sleep. Barbiturates and benzodiazepines extend this intermediate stage at the expense of paradoxical sleep while concomitantly barbiturates suppress the pontine reticular activation characteristic of this sleep stage. During the intermediate stage, thalamocortical responsiveness and thalamic transmission level, which are controlled by brain stem activating influences, are the lowest of all sleep-waking stages. The unusual EEG pattern of this stage is otherwise only observed in the acute intercollicular-transected preparation. Therefore, forebrain structures may be functionally briefly disconnected from the brain-stem during this short-lasting stage, which could possibly account for the mental content of a similar sleep period in humans. In spite of strong evidence in favour of this forebrain deafferentiation hypothesis, other data indicate that the IS is in some way linked either to slow-wave sleep or to paradoxical sleep.

Detection of seven sleep-waking stages in the rat

Seven meaningful sleep-waking stages can be dissociated in the rat. 1) Waking with theta activity in the dorsal hippocampus which corresponds to attentive and/or psychomotor active behavior. 2) Waking without theta activity during which the animal is mainly quiet. 3) The first sleep stage is characterized by cortical slow waves of progressive increasing amplitude. 4) As synchronized sleep deepens, anterior cortex spindles of progressively increasing number, amplitude and duration appear. 5) Just prior to paradoxical sleep occurs an intermediate stage characterized by cortical high amplitude spindles and low frequency theta rhythm. It corresponds to a functional cerveau isolé-like preparation since it is related to a massive decrease of thalamic sensory transmission processes, and acute intercollicular transections induce for hours the same unusual association of EEG patterns. This stage is massively extended at the expense of paradoxical sleep by several psychotropic drugs. 6) Paradoxical sleep without eye movements. 7) Eye movement periods of paradoxical sleep. The central responsiveness and neurophysiological correlations of these stages are discussed.

Study of the 5-HT2 antagonist ritanserin on sleep-walking cycle in the rat

Ritanserin, a 5-HT2 receptor antagonist, was injected intraperitoneally to rats at light onset. It was found that 0.63 mg/kg decreased waking, increased the slow waves characteristic of the first stage of sleep, and decreased paradoxical sleep (PS) during the first four hours. Active waking was further decreased and slow wave stage increased during the following four hours. The number of synchronized and paradoxical sleep phases decreased whereas their duration increased during the first four hours. Ritanserin at 2.5 mg/kg decreased active waking and PS, whereas quite waking and slow wave stage were increased during the first four hours. Quiet waking was increased during the following four hours. It is concluded that serotonin acting on 5 HT2 receptors is actively involved in sleep-waking regulation.