Scoring of sleep and wakefulness by behavioral analysis from video recordings in rhesus monkeys: comparison with conventional EEG analysis

Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

Scoring of sleep and wakefulness by behavioral analysis from video recordings in rhesus monkeys: comparison with conventional EEG analysis E. Balzamo a ,*, P. Van Beers b, D. Lagarde b a

URA 1630 CNRS de Physiopathologie Respiratoire Cellulaire et Inte´gre´e, Institut Jean Roche, Faculte´ de Me´decine Nord, Bd. P. Dramard, 13916 Marseille Ce´dex 20, France b De´partement de Neurophysiologie Ae´rospatiale, IMASSA-CERMA, Bre´tigny sur Orge, France Accepted for publication: 7 April 1997

Abstract Extensive work on sleep-wake cycles in non-human primates has been carried out using conventional EEG scoring. In this study, simultaneous somnopolygrams and video recordings at 1 frame/s were performed on 6 adult rhesus monkeys (Macaca mulatta) during a 24 h period. Wakefulness, NREM sleep and REM sleep were scored by analysis of animal behavior from video data, using characteristic criteria for each state of vigilance. Results were then compared with those of conventional EEG scoring. Values of the total amount for each state obtained by the two scoring methods during the light and the dark periods were significantly closely related (P , 0.001) with a high correlation coefficient for wakefulness (r1 = 0.99956), for NREM sleep (r1 = 0.99641) and for REM sleep (r1 = 0.98708). Moreover, the epoch by epoch analysis between both methods showed a high concordance with percent agreement values of 95.68% for wakefulness, 93.52% for NREM sleep and 94.02% for REM sleep. The number of REM sleep episodes was similarly defined. The patterns of successive sleep-wake cycles determined from both scorings were superimposable, as were the frequent state changes for the same time segments. The video method’s main limitation was that the 4 stages of NREM sleep could not be differentiated. Reliability and advantages of sleep-wake scoring by behavioral analysis are discussed. These results suggest that the video methodology is relevant as a non-invasive technique complementary to conventional EEG analysis for sleep studies in rhesus monkeys.  1998 Elsevier Science Ireland Ltd. Keywords: Sleep-wake scoring; Behavioral analysis; Video; Somnopolygram; Methodology; Rhesus monkey

1. Introduction Extensive work during the early 1960s and later has been carried out into the natural sleep of various animal species from both phylogenetic and ontogenetic aspects. The use of polygraphic recording of sleep and wakefulness was the method of research common to most laboratories for neurophysiological studies or for pharmacological evaluation of the effectiveness of drugs acting on sleep or wakefulness. Visual and manual scoring and quantification of the 3 states of vigilance from polygraphic recordings are time-consuming and a limitation in long-term studies. In order to overcome such problems, different real-time data acquisition and automatic scoring systems have been developed and applied in man (Ferri et al., 1989; Hoelscher et al., 1989;

* Corresponding author. Tel: +33 491 698924; fax: +33 491 698927.

0013-4694/98/$19.00  1998 Elsevier Science Ireland Ltd. All rights reserved PII S(97 )0 0152-1

Kubicki et al., 1989; Salinsky et al., 1992). Results obtained from automatic scoring systems (Oxford Medilog) compared with those obtained with visual conventional EEG analysis were reliable and accurate, with a small deviation and a high correlation (r . 0.9) for all sleep stages, with the exception of brief awakenings (r = 0.69) (Hoelscher et al., 1989). Different conclusions came from the studies of Ferri et al. (1989) and Kubicki et al. (1989). These authors observed a 26.9% difference and pointed out that it was related to the fact that the automatic scoring systems presented difficulties in discriminating rapid stages of sleep (stage 1 and REM) from wakefulness as well as stage 3 from stage 4. The use of closed circuit television has proved useful in observing animal behavior during sleep (NREM sleep) and paradoxical sleep (REM sleep) (Weitzman et al., 1965; Kripke et al., 1968; Sastre and Jouvet, 1979). More recently, using the combination of polygraphic and video recordings,

EEG 96669

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

several studies have been carried out in Macaca mulatta to observe their social behavior during diurnal sleep-wake cycles (Balzamo, 1985), their behavioral activity after administration of a potent psychostimulant substance (Lagarde and Milhaud, 1990) and their natural sleep postures (Lagarde et al., 1992). No quantification of the 3 states of vigilance by behavioral analysis from video recordings was undertaken in these studies. The video technique, rather than being used solely for the observation of animal behavior, could offer also a real interest for sleep-wake scoring, mainly in long-term studies, by means of specific behavioral criteria defined for each state of vigilance. The behavioral analysis presents some limits. They concern the inability to distinguish the NREM sleep stages 1–4 from the animal’s sleeping behavior and to demonstrate the precise state of vigilance such as conventional EEG analysis can do on a moment by moment basis. However, this video methodology does appear to have promise for ascertaining the health status and the true behavior of the animal by a rapid overview of the whole experiment and it also allows the evaluation of the 3 states of vigilance. More importantly, it avoids the need for surgery and the implantation of electrodes. The results obtained by analysis of videotaped behavior must first be evaluated carefully in comparison with those provided by conventional EEG analysis in order to determine the reliability of this system. The aims of the present report were: first, to compare scoring from EEG with that from video analysis in a group of rhesus monkeys during 24 h simultaneous polygraphic and behavior recordings, and second, to validate this methodology and demonstrate its usefulness in complementing any EEG sleep study in this monkey species.

2. Materials and methods 2.1. Subjects Experiments were performed on 6 adult male rhesus monkeys (Macaca mulatta), mean weight of 11.9 ± 1.2 kg. The animals were kept free-ranging in individual cages in an animal facility. Experimental conditions were in agreement with EEC recommendations (November 24, 1986) according to the official European guidelines (86/6091CEE) published in the Journal of European Communities (L. 358, December 18, 1986) and with French regulations (decree no. 87-848 of October 19, 1987). 2.2. Bio-instrumentation Each surgical intervention was performed under total aseptic conditions, in a sterilized room, 2–3 months prior to the experiments. In order to obtain simultaneously the electroencephalogram (EEG), the electrooculogram (EOG) and the electromyogram (EMG) signals, long-term electro-

207

des were implanted according to standardized techniques used in monkeys. Superficial cortical electrodes made of silver points were fixed symmetrically on the dura over the frontal, parietal and suboccipital areas on each side of the sagittal suture. EEG, EOG and EMG electrodes were soldered to a female connector fixed to the skull by steel wires and an acrylic resin prosthesis (Texton). 2.3. Experimental conditions During the experiments, the chronically implanted rhesus monkeys were restrained in special chairs that allowed them to sleep in a position close to their physiological sleeping posture (Milhaud et al., 1980). They were placed, in pairs, side by side in a ventilated and soundproofed isolation room where the ambient temperature was maintained at 21 ± 1°C and the light-dark (LD) cycle was LD 16:8 (light period from 0700 to 2300 h; dark period from 2300 to 0700 h), conditions identical to those of the animal facility. Lighting in the cabin was provided by an incandescent bulb during the light period. Twenty-four hour infrared illumination was also provided. Food was given to the animals and urine and feces removed twice a day, between 0900 h and 1000 h and again between 1800 h and 1900 h during the light period. 2.4. Equipment and recordings Polygraphic parameters were represented by 3 EEG channels, one EOG channel and one EMG channel. The somnopolygrams were recorded continuously throughout the experiment and taped on magnetic supports (AMPEX Model PR 2230, FM, 14 channels; TEAC Model CT 90II and Model HR 30 J). Off-line the taped data was transferred to a paper support for visual EEG evaluation, at a speed of 10 mm/s by means of a polygraphic recorder (VICKERS Medical, 21 channels). An IRIG time code was also printed out. The behavior of each pair of monkeys was recorded continuously at 1 frame/s on a VHS video tape recorder (HITACHI Model EX VT-L30 ED), using an infrared video camera (HITACHI Model KP 140 E/K, CCD 1/2 inch camera). The infrared light source allowed observation and recording of the animals in the dark. The time (hour, minute and second) was inlaid on each frame. The time-clocks for both EEG and video recordings were synchronized. Variations between the two clocks were around 2 s over 24 h. With this video system at 1 frame/s, a 24 h recording lasted only 1 h on video tape, i.e. 1 min of behavior corresponded to 2.5 s on tape at normal speed. This specific video technique gave a good quality display and allowed each behavioral state of vigilance to be well differentiated. 2.5. Procedure Prior to the experiments, the animals were habituated to

208

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

the new conditions for at least 2 full days. Then, the EEG, EOG and EMG signals and the behavioral activity of both animals were recorded simultaneously throughout a 24 h period (0700 h to 0700 h) except for the two intervention periods (0900–1000 h and 1800–1900 h). 2.6. Data analysis Based upon previous studies with the combination of EEG and video recordings (Balzamo, 1985; Lagarde and Milhaud, 1990; Lagarde et al., 1992) or using similar video system (Le Menn et al., 1994), the 3 states of vigilance were easily codified from the video recordings, using characteristic criteria of the monkey’s behavioral activity already described by Weitzman et al. (1965) and Kripke et al. (1968); wakefulness: alert, eyes open or quiet, eyes closed, head resting on the shelf and/or its arms with frequent moving of the head and body; NREM sleep and drowsiness: eyes closed, lack of movement; REM sleep: eyes closed, fall of head muscular tone, movements of the eyes, twitches of face and/or limb extremities, bursts of rapid movements of the external ears. For behavioral analysis, wakefulness, NREM sleep and REM sleep were scored in 1 min epochs, taking into account the video sampling rate of 1 frame/s. A 1 min epoch was the minimum time but also the most satisfactory epoch duration for the quantification of the behavior displayed on the screen. Moreover, the speed of the display could be slowed down in order to detail more accurately the state of vigilance and its duration when any doubt occurred in the observation at normal speed. The method of scoring EEG recordings was based upon those described by Weitzman et al. (1965), Kripke et al. (1968) and Bert et al. (1970) for normal sleep in Macaca mulatta. In the present study, the somnopolygrams were analyzed by 30 s epochs and we considered stages 1, 2, 3 and 4 as NREM sleep. For useful comparison between both methodologies, the EEG analysis data were further expressed in 1 min epochs. In cases where two adjacent 30 s epochs had different sleep-wake stages, we reviewed the original EEG data. The stage which constituted the majority of that minute was then assigned for the entire 1 min epoch. Analysis of behavioral activities and somnopolygrams of each animal was performed in such a way that the data from any one animal were not analyzed consecutively. The EEG and video recordings were analyzed in two time segments (light and dark periods) for each day, in order to define values of the 3 states of vigilance during the day (lights on) and the night, as is conventional for sleep studies. The total amount of time spent in each state was expressed in minutes. Besides quantitative data analysis, in order to determine an exact measure of the percent agreement between the EEG and video methods, we carried out an epoch by epoch analysis, using PC-developed software based upon

the same number of 1 min epochs. Conventional EEG analysis was considered as the reference methodology. In this respect, for tabulated or contingency table data, the rows correspond to the EEG analysis and the columns correspond to the video analysis. In this way, 3 × 3 contingency tables were obtained. Thus, each minute of each state of vigilance would end up in one of the 9 cells. Tables relating to each animal for the light plus the dark period described the distribution of the number of epochs in each state. Cells on the diagonal (upper left to lower right) indicated agreement between the EEG and the video techniques for each state of vigilance. Others cells indicated the number of incorrectly scored epochs by the video analysis. Moreover, a final table relating to all animals for all light plus dark period recordings was obtained by automatic completion. In the same manner, the data were also expressed in percentages. 2.7. Statistics Statistical analysis was carried out using the Intraclass Correlation, a concept generally approached by means of an analysis of variance (ANOVA I) (Zar, 1984). This procedure also assumed that the population variances were equal, aside from assuming random sampling from a bivariate normal distribution. Thus, the two estimated values obtained by conventional EEG analysis and by behavior analysis (Video), i.e. the total amount of time spent in each of the 3 states, expressed in minutes, which corresponded to the variables used, were paired in all subjects during the light period plus the dark period for wakefulness as well as for NREM sleep and for REM sleep. The degree of identity between both quantitative data was evaluated. For this procedure, when the measurements are equal for each pair, the intraclass correlation coefficient (r1) is equal to 1, indicating a perfect positive correlation. The more (r1) approaches 1 in value, the more both values of each pair are closely related. Conversely, if measurements are markedly different, then (r1) is decreased.

3. Results 3.1. Quantitative data Fig. 1 shows a close one-to-one correspondence between the values from both types of analysis (EEG and Video) with a very high correlation coefficient, ascertained by the intraclass correlation, for wakefulness (r1 = 0.99956), for NREM sleep (r1 = 0.99641) as well as for REM sleep (r1 = 0.98708). Values of total duration for each state obtained by the two methodologies were significantly (P , 0.001) closely related (see Table 1). 3.1.1. Light period During this part of the nycthemere, wakefulness as well

209

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

as drowsiness occurred essentially, sleep episodes appearing just before the night. In comparing any one individual’s values, durations of wakefulness obtained by behavior analysis (Video) tended to be slightly lower than those obtained with conventional EEG analysis. This was the case for all monkeys except M.m. O3: from 2 to 17 min, to the benefit of NREM sleep, over 840 min of recording duration (Table 1). These differences were, however, minimized with regard to the mean (±SD) of wakefulness duration for the whole group of 6 monkeys: 677 ± 30 min (Video) vs. 683 ± 81 min (EEG). For REM sleep, values were the same, except in M.m. W4 (15 min vs. 6 min).

Table 1 Durations of states of vigilance from EEG and Video analysis M.m.

Light W4 O2 O3 Z3 F1 F2 Dark W4 O2 O3 Z3 F1 F2

Wake

Sleep

REM

EEG

Video*

EEG

Video*

EEG

Video

676 764 727 757 602 572

673 755 731 740 600 564

158 76 113 83 233 251

152 85 109 100 236 259

6 0 0 0 5 17

15 0 0 0 4 17

124 74 92 140 84 87

130 83 73 137 79 81

299 346 327 304 324 356

291 344 344 303 317 362

57 60 61 36 72 37

59 53 63 40 84 37

Values are expressed in minutes. M.m., Macaca mulatta; Sleep, NREM sleep; REM, REM sleep. *Significance level of intraclass correlation for the 3 states during light and dark periods: P , 0.001.

3.1.2. Dark period It is noted that total duration values obtained by behavioral analysis (Video) were lower for wakefulness, 3–19 min (in monkeys O3; Z3; F1; F2), as well as for NREM sleep, 1–8 min (in monkeys W4; O2; Z3; F1), but higher for REM sleep from 2 to 12 min (in monkeys W4; O3; Z3; F1), over 480 min of recording duration. However, the means (±SD) of duration of each state for the group of 6 monkeys obtained by both methodologies were quite similar: wakefulness, 97 ± 28 min (Video) vs. 100 ± 26 min (EEG); NREM sleep, 327 ± 27 min vs. 326 ± 23 min; and REM sleep, 56 ± 17 min vs. 54 ± 14 min. 3.1.3. Light plus dark period Only slight differences were also observed for the means (±SD) of duration relating to the light plus dark period: wakefulness, 774 ± 92 min (Video) vs. 783 ± 92 min (EEG); NREM sleep, 484 ± 85 min vs. 478 ± 85 min; and REM sleep, 62 ± 17 min vs. 59 ± 13 min. 3.2. Epoch by epoch analysis

Fig. 1. Values of wakefulness (Wake), NREM deep (Sleep) and REM sleep (REM) obtained by conventional analysis (EEG) in relation to values obtained by behavior analysis (Video). Note: min = minute.

Because of the high number of tables, we report here only the main results of the epoch by epoch analysis, relating to the number of epochs and the percentage for all light plus dark period recordings in all animals (Table 2). The percent agreement between both methodologies relative to wakefulness reached the maximum value: 95.68%. Out of a total of 4699 epochs scored for wakefulness by EEG analysis, 4496 epochs were correctly scored by behavioral analysis. Most of the differently scored wakefulness epochs were allocated to NREM sleep, i.e. 197 epochs (4.19%). Only 6 wakefulness epochs (0.13%) were scored for REM sleep. Of a total of 2870 NREM sleep epochs, 2684

210

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

Table 2 Epoch by epoch analysis for each state of vigilance showing the number (n) and the corresponding percentage (%) of epochs scored in agreement (diagonal from left to right) between EEG (rows) and Video (columns) methods and of those epochs scored differently by the video analysis State

Wake Sleep REM

n

%

Wake

Sleep

REM

Wake

Sleep

REM

4496 150 0

197 2684 21

6 36 330

95.68 5.23 0.0

4.19 93.52 5.98

0.13 1.25 94.02

epochs (93.52%) were identically scored. Of the remaining discrepantly scored NREM sleep epochs, 150 epochs (5.23%) were allocated to wakefulness and 36 epochs (1.25%) to REM sleep. The value of percent agreement between the two techniques for REM sleep was 94.02% (330 epochs of a total of 351). The 21 incorrectly scored epochs were allocated to NREM sleep. 3.3. Sleep-wakefulness cycles The patterns of successive episodes of wakefulness, NREM sleep (drowsiness or sleep) and REM sleep that occurred during the light or dark periods, scored by EEG and video analysis, were compared in each of the 6 rhesus monkeys. We observed that the patterns obtained by both methodologies were superimposable for the same time segments, as illustrated in Fig. 2. Sleep cycles, as well as the frequent state change, were reproduced well in the video analysis. Nevertheless, in some cases the behavioral analysis discrimination appeared less precise such as in the case of brief drowsiness or NREM sleep episodes observed on the somnopolygrams, which were then scored as wakefulness by video analysis, and vice versa. Concerning REM sleep, the number of episodes, 8–12 depending on the animal, was similarly defined by the two techniques, except for one episode of 1 min duration which was identified only on the EEG recordings (Fig. 2, at 0215 h). The percentage of agreement for an equivalent duration of REM sleep episodes defined by both methodologies in the 6 monkeys, averaged 62% of the total. For the remainder, the durations scored by video analysis were mainly, in equal number, of 1 min either less or more than durations obtained by EEG analysis.

activities of the monkeys. The shortening in time gave a rapid overview of the sequences of the sleep-wake cycles. Based upon the criteria described in the methods section, the 3 states of vigilance were able to be correctly evaluated, as they occurred. A further advantage is that a 24 h video recording can be scored within 2 h. However, conventional EEG analysis must be considered as the reference methodology. The analysis of the somnopolygrams allowed, at any moment, a precise definition of the state of vigilance whatever the behavioral activities of the monkey. In the present study, the analysis of videotaped behavior appeared to constitute a reliable methodology for scoring sleep and wakefulness, showing highly significant, closely related results with a high correlation coefficient in comparison with those obtained by conventional EEG analysis. As was described in the results, pairs of values of total duration, whatever the animal, as well as the mean durations for each state of vigilance, during the light, the dark and the light plus dark period, were of the same order from both types of analysis. However, differences relating to the mean or individual values originating from video analysis were observed. These concerned: wakefulness, which tended to be slightly underestimated (9 min for the mean values relating to the light plus dark period) and less than a maximum of 17 min over 840 min of recording during the light period, i.e. ,2% of variation, to the benefit of NREM sleep; REM sleep, which tended to be overestimated by 3 min at the expense of NREM sleep, mainly at night. The epoch by epoch analysis which gives an exact mea-

4. Discussion It must be underlined from the results of this study that data obtained using this particular system of video recordings at 1 frame/s, in artificial or infrared light, gave the same quality of display as full frame video recordings. Surprisingly, the accelerated motion, with the tape playing at normal speed, gave an excellent caricature of the behavioral

Fig. 2. Hypnograms from data obtained by conventional analysis (EEG) and behavior analysis (Video) during light and dark periods in M.m. O2 rhesus monkey. W, wakefulness; S, NREM sleep; REM, REM sleep.

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

sure of the percent agreement between the two methods tended to confirm the reliability of the video analysis. Results showed a high value for the percent agreement, quite similar for NREM sleep and REM sleep (93.52% and 94.02%, respectively) and maximal for wakefulness (95.68%). However, it turned out that the behavioral analysis did not easily identify some of the wakefulness episodes and allocated them, incorrectly, to NREM sleep (197 epochs, i.e. 4.19%). Also, some epochs of NREM sleep episodes were interpreted as wakefulness (150 epochs, i.e. 5.23%) or as REM sleep (36 epochs, i.e. 1.25%). Concerning REM sleep, the behavioral analysis scored only 21 epochs (5.98%) as NREM sleep differently from the EEG analysis. Looking at the differences between video and EEG scoring, we can consider them as acceptable with regard to the variability of results between two readers or between different groups of readers in conventional scoring of the same EEG sleep recordings (Ferri et al., 1989). Besides, the patterns of sleep-wake cycles, occurring successively throughout both the light and dark periods, matched for both types of analysis and the frequent state changes were well reproduced. On the other hand, the reliable scoring of sleep and wakefulness obtained by video analysis should be seen with reference to that provided by automatic analysis using the Oxford Medilog 9000 System (Ferri et al., 1989; Hoelscher et al., 1989) or quantitative EEG frequency analysis in man (Salinsky et al., 1992) and in rat (Vivaldi et al., 1984) when compared with conventional EEG scoring. How can these differences be explained? Wakefulness episodes were not identified by video analysis when the behavioral attitude of the monkey on the screen display was quiet, without head movements and eyes closed, resembling drowsiness or a sleep posture. They concerned short duration periods during quiet wakefulness or in the transient waking state during NREM sleep. Conversely, some short drowsiness episodes or moments at the very beginning of sleep onset were not easily identified and interpreted as wake. Because these events were one of the characteristics of the sleep-wake patterns of monkeys, their repetitive number throughout the nycthemere leads to incorrectly scored epochs, 197 and 150, respectively. The amount of REM sleep presented a significant identity in the comparison between video and EEG analysis as well as similar values for the mean of the whole group. Even though these data seemed satisfactory, REM sleep scoring by video needed to be examined more closely. In some cases, short NREM sleep epochs preceding or within a REM sleep episode, as shown by EEG analysis, were interpreted as REM sleep by behavioral analysis. Also, some epochs of REM sleep were not recognized and were scored as NREM sleep. These discrepancies might be related to the occurrence of bursts of rapid eye movements preceding the decrease of voltage of slow EEG activities and also to the proper expression of REM sleep where epochs of REM sleep were accompanied by few rapid eye movements of

211

low amplitude and where transient stage 1 NREM sleep occurred within an episode. The first event has been previously described in rhesus monkeys (Weitzman et al., 1965) and in cats (Thomas and Benoit, 1967). The second event has been commonly observed in non-human primates and in man. However, the percentages of non-agreement with respect to REM sleep remained minor in comparison with those of wakefulness and NREM sleep. Thus, the use of video analysis for scoring obviously presents certain limitations. The video system records only attitude, position and movements of the head and/or body of the monkey. The inability to distinguish the 4 stages of NREM sleep from the sleeping posture of the animal constitutes the main limitation of video analysis for all sleep studies. On the other hand, this system is not capable of distinguishing between stages of wake and 1, as are the EEG recordings which show distinct cortical activities, because, at various moments, the animal’s behavior is similar for quiet wakefulness and sleepiness. This remained the main scoring problem in using behavior analysis. Similar problems have occurred with automatic analysis in man, as pointed out by Ferri et al. (1989) and Kubicki et al. (1989).To date, no advance in video techniques can be applied in order to reveal sleepiness and to improve the scoring. Although these factors cannot be discarded, several advantages of this methodology should be pointed out. The video recordings brought to the somnopolygrams information about the behavioral activities of the monkeys during sleep-wakefulness rhythm and allowed detailed analysis of events throughout the experiment. At any one moment, the behavior and the health status of the rhesus monkeys were known and could explain the disturbances of the sleepwake pattern noticed on the EEG recordings. The present video system at 1 frame/s allowed also the acquisition of data that required fewer video tapes, which in long term studies lowers costs and time spent in EEG recording and scoring. These advantages seemed thus to bring a positive answer to the related considerations pointed out by Ferri et al. (1989). A rapid overview of a whole experiment is possible which is of great interest for long-term pharmacological studies which can thus be carried out with continuous recordings. With regard to the increasing concern for the well-being of experimental animals, especially non-human primates, the use of this method prevents surgery and the implantation of electrodes. This can also offer the opportunity for studying sleep-wake cycles in particular conditions where polygraphic recordings would not be possible. The technique of video analysis has contributed to numerous applications in medical clinics. Somnopolygrams and videotaped recordings as well as split-screen video-EEG polysomnography (VPSG) have been intensively used for inpatient monitoring or diagnostics of various psychiatric or neurologic sleep disorders (Pierelli et al., 1989; Aldrich and Jahnke, 1991; Kempenaers et al., 1994; Zucconi et al., 1995). These reports have shown that VPSG was superior

212

E. Balzamo et al. / Electroencephalography and clinical Neurophysiology 106 (1998) 206–212

to standard polysomnography for evaluation or diagnosis and may also be a suitable alternative to intensive inpatient monitoring. The increasing development of computerized digitalizing of the analogic video data with full frame or 1 frame/s will undoubtedly increase the utilization of this tool in clinical research. With respect to the above mentioned aspects and to the results reported in this study, the use of behavioral analysis as a methodology for scoring sleep and wakefulness from video data may be proposed for further research in sleepwake cycles on rhesus monkeys. The present results have shown the efficiency and the reliability of sleep-wake cycles scoring by behavioral analysis and validated the behavioral criteria and the technique. This methodology must be considered, besides the automatic analysis largely used at the present time, as a useful and a complementary technique to conventional EEG analysis for sleep studies in non-human primates.

Acknowledgements The authors wish to thank Mr. Henri Burnet for his help in the statistical analysis and Dr. Simon Thornton for revision of the English. This work was funded by grants from the Centre National d’Etudes Spatiales (C.N.E.S.94/250; 95/296).

References Aldrich, M.S. and Jahnke, B. Diagnostic value of video-EEG polysomnography. Neurology, 1991, 41: 1060–1066. Balzamo, E. Comportement social de couples de macaques rhe´sus, place´s en contention pour des expe´rimentations neurophysiologiques et neuropharmacologiques. Sci. Tech. Anim. Lab., 1985, 10: 91–95. Bert, J., Pegram, V., Rhodes, J.M., Balzamo, E. and Naquet, R. A comparative sleep study of two Cercopithecinae. Electroenceph. clin. Neurophysiol., 1970, 28: 32–40. Ferri, R., Ferri, P., Colognola, M., Petrella, M.A., Musumeci, S.A. and Bergonzi, P. Comparison between the results of an automatic and a visual scoring of sleep EEG recordings. Sleep, 1989, 12: 354–362.

Hoelscher, T.J., McCall, W.V., Powell, J., Marsh, G.R. and Erwin, C.W. Two methods of scoring sleep with the Oxford Medilog 9000: comparison to conventional paper scoring. Sleep, 1989, 12: 133–139. Kempenaers, C., Bouillon, E. and Mendlewicz, J. A rhythmic movement disorder in REM sleep: a case report. Sleep, 1994, 17: 274–279. Kripke, D.F., Reite, M.L., Pegram, V., Stephens, L.M. and Lewis, O.F. Nocturnal sleep in rhesus monkeys. Electroenceph. clin. Neurophysiol., 1968, 24: 582–586. Kubicki, St., Ho¨ller, L., Berg, I., Pastelak-Price, C. and Dorow, R. Sleep EEG evaluation: comparison of results obtained by visual scoring and automatic analysis with the Oxford sleep stager. Sleep, 1989, 12: 140– 149. Lagarde, D. and Milhaud, C. Electroencephalographic effects of modafinil, an alpha-1-adrenergic psychostimulant, on the sleep of rhesus monkeys. Sleep, 1990, 13: 441–448. Lagarde, D., Anton, G., Bonnin, F. and Milhaud, C. Behaviour of the laboratory rhesus Macaque during nocturnal sleep. Folia Primatol., 1992, 58: 161–175. Le Menn, M., Lagarde, D., Van Beers, P., Plagnes, D., Sarifian, D. and Serra, A. Actographie par traitement d’images: inte´reˆt dans l’analyse comportementale des Primates et les e´tudes de pharmacologie. Sci. Tech. Anim. Lab., 1994, 19: 215–222. Milhaud, C., Klein, M. and Merkel, M. A new restraining chair for rhesus monkeys (Macaca mulatta). J. Med. Primatol., 1980, 9: 62–70. Pierelli, F., Chatrian, G.E., Erdly, W.W. and Swanson, P.D. Long-term EEG-video-audio monitoring: detection of partial epileptic seizures and psychogenic episodes by 24-hour EEG record review. Epilepsia, 1989, 30: 513–523. Salinsky, M.C., Oken, B.S., Kramer, R.E. and Morehead, L. A comparison of quantitative EEG frequency analysis and conventional EEG in patients with focal brain lesions. Electroenceph. clin. Neurophysiol., 1992, 83: 358–366. Sastre, J.P. and Jouvet, M. Comportement onirique du chat. Physiol. Behav., 1979, 22: 979–989. Thomas, J. and Benoit, O. Individualisation d’un sommeil a` ondes lentes et activite´s phasiques. Brain Res., 1967, 5: 221–235. Vivaldi, E.A., Pastel, R.H., Fernstrom, J.D. and Hobson, J.A. Long term stability of rat sleep quantified by microcomputer analysis. Electroenceph. clin. Neurophysiol., 1984, 58: 253–265. Weitzman, E.D., Kripke, D.F., Pollak, C. and Dominguez, J. Cyclic activity in sleep of Macaca mulatta. Arch. Neurol. (Chic.), 1965, 12: 463– 467. Zar, J.H. Biostatistical Analysis. Prentice Hall International Editions, Englewood Cliffs, NJ, 1984. Zucconi, M., Oldani, A., Perini-Strambi, L. and Smirne, S. Arousal fluctuations in non-rapid eye movement parasomnias: the role of cyclic alternating pattern as a measure of sleep instability. J. Clin. Neurophysiol., 1995, 12: 147–154.