Académique Documents
Professionnel Documents
Culture Documents
8, AUGUST 2005
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
AGARWAL et al.: DETECTION OF REMs IN SLEEP STUDIES 1391
criterion and have time duration of 0.33 to 1 s (1–3 Hz band- the fourth REM periods were used for the detection analysis. In
pass filter). Padovan and Pansini [10] showed the advantage of the cases where there was no clear fourth REM period, the third
a three channel montage in determining the direction of EMs. REM period was used. All NREM epochs within each 15-min
Degler et al. [6] used the techniques of [5] and [10] to deter- REM periods were excluded from analysis. Independently from
mine the direction of EMs. Ktonas and Smith [7] modified the any algorithm development, an experienced sleep technologist
techniques of [5] and [6] and analyzed a random selection of (one of the authors) manually scored the presence of REM events.
stage REM epochs in adult PSGs. Ninety-one of ninety-seven The only clear criterion that was used in the manual scoring was
manually marked REMs were correctly identified. Doman et that no two REM events be closer than 0.5 s.
al. [11], presented a method based on amplitude and slope of
deflection and showed acceptable performance using ten ran- B. REM Detection Methods
domly selected 1-min REM sleep epochs. However, no details We present a method of automatic detection of fast horizontal
were provided as to how the manual and the automatically de- ocular activity in any sleep stage. Subsequent processing can
tected EMs were matched. Minard and Krausman [12] presented subclassify the detections into various sleep stages to isolate
a detector using the high speed leading edge and out-of-phase only those that occur in the REM stage. The method attempts
characteristics of REMs (degree of synchrony between left and to incorporate the various EM features that have been used in
right outer-canthi recordings could be specified). The detector some previously published techniques. Two salient features of
did not use a REM amplitude criterion; however, tuning was the horizontal REMs are the rapid change in amplitude (fast de-
required for different recording settings. No analysis was pre- flections from baseline) and the phase-reversed synchrony in the
sented. McPartland and Kupfer [13] developed a detector using left and right EOG channels. Thus, the proposed method uses
a minimum amplitude criterion and the requirement that the two EOG channels (LOC and ROC) with the underlying ratio-
two EOG channels (left and right) have conjugate movements nale that the REMs must occur in synchrony in the two channels,
(synchrony with opposite polarity within 100 ms of each other). be phase-reversed and be of sufficient amplitude. Detections are
There have been other methods presented in the literature, but made on a block-by-block basis where the duration of the de-
most are variations of the above. More recently, a matched fil- tection block is the same as the duration of the staging epoch.
tering technique [9] based on morphology of neonate REMs has The basic principle in this detection method is to generate a set
been presented. This approach requires the preselection of REM of candidate REMs (CREMs) that can be subsequently filtered
templates. It is well known that the matched filters (MFs) are such that only those that have characteristic properties identi-
optimal in the presence of white noise; however, the presence fying REMs are retained. The following describes the different
of artifacts can degrade their performance. Using a Chi-square components of the algorithm.
analysis, an overall accuracy of 72.1% ( , ) 1) Data Preprocessing: The EMs are generally of two
was reported, though no discussion was presented describing types: 1) the slow components known as the slow-eye move-
how computer and the visually detected REMs were matched ments (SEMs) and 2) fast components know as the REMs. To
or how the templates were selected. separate the REMs from SEMs, prior to any detection, the two
channels of EOG data are filtered using a fourth-order digital
II. METHODS Butterworth bandpass filter with cutoffs at 1 and 5 Hz to yield
the filtered EOG data, and . This effectively
A. Subjects: Data Acquisition minimizes false detections due to SEMs as REMs as well as any
In the development and the analysis of the REM detection high frequency noise. It is possible that the use of a relatively
method, 10 complete night-sleep recordings (total of 82.5 h) narrowband filter may cause ringing due to impulsive noise,
were randomly selected from an existing set previously recorded and hence cause further false detections. However, by using
(approximately 8–10 yr prior to this study) with the Stellate additional properties of REMs (effectively several layers of
(Montreal, Canada) acquisition system at a 128-Hz sampling artifact rejection), this can be minimized. In our data analysis,
rate and calibrated to 7.5 . The 10 recordings were from we did not see this to be an issue.
different subjects (seven males, three females) ranging from 14 to 2) Candidate REM Detection: The REM detection criterion
60 yr in age ( , ). The only inclusion crite- is generated as the negative instantaneous product of the two
rion was that the recordings included two EOG channels (left and filtered EOG sequences
right EOGs) and REM periods. The EOG channel derivation was
slightly different from that described in the introduction—LOC (1.1)
and ROC were referenced to the ipsilateral ear. No record of
data acquisition filters was retained; however, the laboratory where n is the discrete time index. To remove all negative
routinely uses second-order filters with highpass cutoff at 0.1 peaks (in-phase activity) and the baseline noise around zero, a
Hz and a lowpass cutoff of 35 Hz. We expect this was the case threshold is applied to such that
for the considered EOG data. All recordings were staged using
20-s epochs according to the Rechtschaffen and Kales sleep (1.2)
staging criteria [14] prior to any development or analysis. The
10 recordings were randomly split into two groups: five for the Fig. 2 illustrates the detection of CREMs and the derivation
development of the REM detection method and five for the testing of using a 20-s staging epoch. A peak detection method is
of the method. In all recordings, the first 15 min of the second and applied to to generate candidate REMs (CREMs) using a
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
1392 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 52, NO. 8, AUGUST 2005
Fig. 2. Detection of CREMs. (a) Left and right EOG signals. (b) Bandpass
filtered signal in (a). (c) Product of bandpass filtered EOGs in (b). (d) Final
CREM detection criterion with the sliding window of 2L around the current
sample. The window in the figure represents 1 s, but has been made larger for
illustration.
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
AGARWAL et al.: DETECTION OF REMs IN SLEEP STUDIES 1393
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
1394 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 52, NO. 8, AUGUST 2005
TABLE I
TESTING DATA RESULTS WITH OPTIMAL PARAMETERS
TABLE II
TRAINING DATA RESULTS WITH OPTIMAL PARAMETERS
TABLE III
TESTING DATA RESULTS AFTER ASSESSMENT OF ERRORS
TABLE IV
PERCENTAGE OF REM EVENT DETECTION IN EACH SLEEP STAGE FOR THE TEST DATA
in sensitivity was observed; however, many of the newly scored scoring of these events is tedious and very time-consuming, a
events were in fact detected by the automatic algorithm as sug- situation that has fueled the development of automatic detec-
gested by an overall increase of 7.5% in specificity. The in- tion methods. However, there are no commercially available
creased count in manual REM count was likely due to increased REM detection packages. Various sleep research laboratories
vigilance as the scorer was aware of the objective. Needless have developed their own REM detector with varying defini-
to say, the interscorer as well as the intrascorer variability can tions of REMs.
be great in the visual identification of individual REM events Many of the reported strategies of REM detection rely of
and can create significant scorer bias. Furthermore, the manual EM models that typically include such features as binocular
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
AGARWAL et al.: DETECTION OF REMs IN SLEEP STUDIES 1395
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.
1396 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 52, NO. 8, AUGUST 2005
application of the method in different laboratories, it is antici- [10] I. Padovan and M. Pansini, “New possibilities of analysis in electronys-
pated that in the initial stages, the user will adjust the sensitivity tagmography,” Acta Otolaryngol., vol. 73, pp. 121–125, 1972.
[11] J. Doman, C. Detka, T. Hoffman, D. Kesicki, J. P. Monahan, D. J.
threshold to meet his/her preference. Buysse, C. F. Reynolds 3rd, P. A. Coble, J. Matzzie, and D. J. Kupfer,
It is difficult to compare the performance to other methods in “Automating the sleep laboratory: implementation and validation of
the literature since no standardized criterion of performance has digital recording and analysis,” Int. J. Biomed. Comput., vol. 38, pp.
277–290, 1995.
been utilized. The reported performance metrics or the methods [12] J. G. Minard and D. Krausman, “Rapid eye movement definition and
to compare manual and automatic scoring are varying or un- count: an on-line detector,” Electroencephalogr. Clin. Neurophysiol.,
clear. Doman et al. [11] showed a 90% sensitivity and 93% vol. 31, pp. 99–102, 1971.
[13] R. J. McPartland and D. J. Kupfer, “Computerised measures of electro-
specificity of REM detections on ten 1-min REM sleep epochs. oculographic activity during sleep,” Int. J. Biomed. Comput., vol. 9, pp.
However, this is a very small data set and the details of how 409–419, 1978.
manually scored and automatically detected events were com- [14] A. Rechtschaffen and A. Kales, A Manual of Standardized Termi-
nology, Techniques and Scoring System for Sleep Stages of Human
pared are not reported. Our method has been evaluated on a rela- Subjects. Los Angeles: Brain Information Service/Brain Research
tively large independent data set and it offers several key advan- Institute, Univ. California, 1968.
tages: 1) the possibility of tailoring the detection sensitivity by
adjusting a single threshold; 2) the method does not explicitly
require amplitude levels or any other REM properties to make Rajeev Agarwal (M’88) received the M.A.Sc. and
Ph.D. degrees, both in electrical engineering, in 1991
detections; 3) epochs containing artifacts are automatically re- and 1995, from Concordia University, Montreal, QC,
jected; 4) the method is robust in minimizing REM detection Canada.
in NREM sleep as described in Table IV; and 5) the method is He spent one year as a Postdoctoral Fellow
with the Montreal Neurological Institute, McGill
computationally very simple and can easily be implemented in University, Montreal, followed by a two-year
real-time. The major advantage of this method is the minimal NSERC-supported Industrial Research Fellowship
a priori information required for REM detection. For example, at Stellate Systems, Montreal. He is currently the
Director of Research and Development at Stellate
a more recently published approach [9] used matched filtering Systems and holds an Adjunct Associate Professor
techniques. In such approaches, it is necessary to use REM tem- appointment with Concordia University. His research interests include biomed-
plates for the detection. The robustness of such a method is re- ical signal processing (particularly, pattern detection in the EEG of epileptic
patients, analysis of polysomnograms, ICU EEG monitoring, and evoked
stricted by the design/selection of the templates. potentials), statistical signal processing, signal detection, and estimation.
V. CONCLUSION
Tomoka Takeuchi received the Ph.D. degree in behavioral neuroscience from
It appears that automatic detection of REMs can be achieved the Division of Engineering, Hiroshima University, Japan, in 1997. In 2002, she
with reasonable performance and with a robust and easily ad- received the chercheur-boursier, junior 1, clinique et épidémiologie from the
justable method. It will be necessary to perform a more exten- Fonds de la recherche en santé du Québec, QC, Canada.
She worked as a Postdoctoral Fellow in Brock University, Ontario, ON,
sive evaluation on a varied subject group, although the important Canada, supported by the Japan Society for the Promotion of Sciences during
lack of a standard definition of REMs may render this validation 1997–2000. In 2001, she moved to Montreal as a Postdoctoral Fellow with the
disputable. Centre d’étude du Sommeil, Hospital du Sacré-Coeur de Montréal. Currently,
she is an Assistant Professor of research, Département de Psychiatrie, Univer-
sité de Montréal, and a researcher with the Centre d’étude du Sommeil, Hôspital
REFERENCES du Sacré-Coeur de Montréal. Her current interest includes mechanisms of
parasomnia dreaming, and individual vulnerability to sleep disturbances.
[1] R. J. McPartland, B. L. Weiss, and D. J. Kupfer, “An objective measure
of REM activity,” Physiolog. Psychol., vol. 2, pp. 441–443, 1974.
[2] F. G. Foster, D. J. Kupfer, P. Coble, and R. J. McPartland, “Rapid eye
movement sleep density. An objective indicator in severe medical-de- Suzie Laroche received the diploma in electrophysiology in 1995 from
pressive syndromes,” Arch. Gen. Psych., vol. 33, pp. 1119–1123, 1976. Ahuntsic College, Montreal, QC, Canada.
[3] R. J. McPartland, D. J. Kupfer, P. Coble, D. H. Shaw, and D. G. Spiker, From 1995 to 1999, she was a Technician with the Polysomnographic (PSG)
“An automated analysis of REM sleep in primary depression,” Biol. Laboratory, Ottawa General Hospital, Ottawa, ON, Canada, and during the last
Psych., vol. 14, pp. 767–776, 1979. three of these years, she took part in teaching courses in PSG at the hospital.
[4] C. Smith and L. Lapp, “Increases in number of REMS and REM density From 1999 to 2003, she was with the Technical Support Team and participated in
in humans following an intensive learning period,” Sleep, vol. 14, pp. the activities of the Research and Development team at Stellate Systems, Mon-
325–330, 1991. treal. She is currently an instructor in electrophysiology at Ahuntsic College.
[5] J. R. Smith, M. J. Cronin, and I. Karacan, “A multichannel hybrid system
for rapid eye movement detection (REM detection),” Comput. Biomed.
Res., vol. 4, pp. 275–290, 1971.
[6] H. E. Degler Jr., J. R. Smith, and F. O. Black, “Automatic detection Jean Gotman received the engineering degree
and resolution of synchronous rapid eye movements,” Comput. Biomed. from Ecole Superieure d’Electricite, Paris, France,
Res., vol. 8, pp. 393–404, 1975. in 1969, the Master’s degree in engineering from
[7] P. Y. Ktonas and J. R. Smith, “Automatic REM detection: modifications Dartmouth College, Hanover, NH, in 1971, and the
on an existing system and preliminary normative data,” Int. J. Biomed. Ph.D. degree in neurological sciences from McGill
Comput., vol. 9, pp. 445–464, 1978. University, Montreal, QC, Canada, in 1976.
[8] J. Martinerie, J. P. Joseph, and M. Naillon, “Computerized detection of He is currently working at the Montreal Neuro-
rapid eye movements during paradoxical sleep,” Int. J. Biomed. Comput., logical Institute and Hospital and is a Professor with
vol. 11, pp. 163–171, 1980. the Departments of Neurology and Biomedical Engi-
[9] G. M. Hatzilabrou, N. Greenberg, R. J. Sclabassi, T. Carroll, R. D. neering of McGill University. He is also the founder
Guthrie, and M. S. Scher, “A comparison of conventional and matched and president of Stellate Systems. His research inter-
filtering techniques for rapid eye movement detection of the newborn,” ests center on signal processing and pattern recognition of the EEG, particularly
IEEE Trans. Biomed. Eng., vol. 41, no. 10, pp. 990–995, Oct. 1994. in the field of epilepsy and of long-term monitoring in the intensive care unit.
Authorized licensed use limited to: IEEE Xplore. Downloaded on April 23, 2009 at 04:13 from IEEE Xplore. Restrictions apply.