Clinical Neurophysiology

Transcription

1 Clinical Neurophysiology 120 (2009) Contents lists available at ScienceDirect Clinical Neurophysiology journal homepage: EEG in the healthy term newborn within 12 hours of birth I. Korotchikova a,b, *, S. Connolly b,c, C.A. Ryan a,b, D.M. Murray a,b, A. Temko b, B.R. Greene b, G.B. Boylan a,b a Department of Paediatrics and Child Health, University College Cork, c/o Clinical Investigation Unit, Cork University Hospital, Wilton, Cork, Ireland b Neonatal Brain Research Group, University College Cork, Cork, Ireland c Department of Clinical Neurophysiology, St. Vincent s University Hospital, Dublin, Ireland article info abstract Article history: Accepted 21 March 2009 Available online 8 May 2009 Keywords: Neonatal EEG Sleep state Spectral edge frequency Spectral entropy Relative delta power Objective: To characterise and quantify the EEG during sleep in healthy newborns in the early newborn period. Methods: Continuous multi-channel video-eeg data was recorded for up to 2 hours in normal newborns within 12 hours of birth. The total amount of active (AS) and quiet sleep (QS) was calculated in the first hour of recording. The EEG signal was quantitatively analysed for symmetry and synchrony. Spectral edge frequency (SEF), spectral entropy (H) and relative delta power (d R ) were calculated for a ten-minute segment of AS and QS in each recording. Paired t-test and Wilcoxon rank sum test were used for data analysis. Results: Thirty normal newborn babies were studied, 10 within 6 hours of birth and 20 between 6 and 12 hours. All babies showed continuous symmetrical and synchronous EEG activity and well-developed sleep wake cycling (SWC) with the median percentage of AS 48.5% and QS 36.6%. Quantitative EEG analysis of sleep epochs showed that SEF and H were significantly higher (p < ) and d R was significantly lower (p < ) in AS than in QS. Conclusion: The normal newborn EEG shows symmetrical and synchronous continuous activity and welldeveloped SWC as early as within the first 6 hours of birth. Quantitative analysis of the EEG in the early postnatal period reveals differences in SEF, H and d R for AS and QS periods. Significance: These findings may have implications for quantitative analysis of the newborn EEG, including the EEG of babies with hypoxic ischaemic encephalopathy. Ó 2009 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. 1. Introduction Very little is known about the EEG patterns of the healthy term newborns in the immediate postnatal period. Much of what we know about neonatal EEG is based on studies recorded in the first weeks of life. Of the few studies that have examined EEG in the first 24 hours of birth, all were limited to neurologically abnormal infants (Pezzani et al., 1986; Pressler et al., 2001). The immediate postnatal period is the time when the EEG is often performed on critically ill neonates. In infants with hypoxic ischaemic encephalopathy (HIE), multi-channel and amplitude integrated EEG are increasingly being used in the first 6 hours after birth to decide eligibility for therapeutic hypothermia (Gluckman et al., 2005; Wyatt et al., 2007). However, reference normative data are not available in the very early postnatal period making interpretation of the abnormal EEG in this period very difficult. * Corresponding author. Address: Department of Paediatrics and Child Health, University College Cork, c/o Clinical Investigation Unit, Cork University Hospital, Wilton, Cork, Ireland. Tel.: ; fax: address: I.Korotchikova@ucc.ie (I. Korotchikova). Neonatal EEG can be assessed qualitatively and quantitatively. Qualitative EEG analysis is mainly used for clinical purposes. It is based on visual interpretation of the EEG signal and describes such background features as amplitude, frequency, continuity of the EEG and sleep wake cycling (SWC). Quantitative EEG analysis is a method predominantly used in research and includes time and frequency domain analysis. Decisions regarding long term prognosis following neonatal HIE are made based on qualitative EEG analysis. Low background amplitude and discontinuity of EEG activity (burst suppression) occur following significant hypoxic-ischaemic injury, and are associated with a poor prognosis (Watanabe et al., 1980; Wertheim et al., 1994; Menache et al., 2002). Disruption of SWC has been described in neonatal post-asphyxial injury, with a decrease in active sleep (AS) and an increase in the proportion of quiet (QS) and indeterminate sleep (IS) (Lombroso, 1985; Scher, 1994; Scher et al., 2002). The absence of SWC soon after birth has been associated with a poor neurodevelopmental outcome (Pezzani et al., 1986). In addition, there is a close association between the early return of SWC and normal neurological outcome (Thorngren- Jerneck et al., 2003; ter Horst et al., 2004; Osredkar et al., 2005) /$36.00 Ó 2009 International Federation of Clinical Neurophysiology. Published by Elsevier Ireland Ltd. All rights reserved. doi: /j.clinph

2 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) Nevertheless, these background EEG features have not been studied systematically in healthy infants in the immediate postnatal period and, in particular, sleep patterns have not been measured or quantified. Another important aspect in the assessment of the neonatal EEG is the presence of interhemispheric symmetry and synchrony. Healthy full-term newborn babies should display symmetrical and synchronous activity of the EEG signal over both hemispheres (Varner et al., 1977, 1978; Lombroso, 1985; Scher, 1996). Adult EEG studies have revealed that quantitative frequency domain measures such as spectral edge frequency (SEF), spectral entropy (H) and relative delta power (d R ) are clinically significant for monitoring sleep states and depth of anaesthesia (Long et al., 1989; Armitage and Roffwarg, 1993; Gurman, 1994; Fell et al., 1996). However, no attempts have been made to apply spectral analysis to the neonatal EEG in the immediate postnatal period to try and quantitatively identify the sleep state in the healthy term newborn baby. The only study that has shown the differences in SEF between AS and QS in healthy infants was performed on the third day of life and included a mixture of term and preterm infants (Bell et al., 1991). Spectral measures such as SEF have also been examined in relation to outcome in preterm infants (Inder et al., 2003). However, this study did not consider the possible changes in SEF with behavioural state. Attempts have also been made to examine the changes in SEF in asphyxiated infants, but have been limited to a comparison of bursts of activity during burst suppression and tracé alternant (Thordstein et al., 2004). Information regarding spectral analysis in the immediate postnatal period in normal infants is not available. The aim of this prospective study was to describe and quantify normative EEG patterns during sleep in healthy term newborns during the first 12 hours after birth. These data will contribute normal reference data for studies of babies with HIE. 2. Subjects and methods 2.1. Subjects The study was performed between October 2005 and August 2007 in a large maternity hospital with approximately 8,000 deliveries per year. Full-term newborn babies were recruited from the postnatal wards in Cork University Maternity Hospital. Babies were enrolled as healthy babies if they met the following criteria: Gestation >37 weeks. No requirement for resuscitation following delivery. Apgar scores of >8 at 5 min. Normal cord ph (>7.1). Exclusion criteria were: Maternal epilepsy or diabetes. Birth weight <2.5 kg. Congenital anomalies. Admission to the neonatal unit for special or intensive care. Following parental consent, babies were examined using the Amiel-Tison assessment, a standardised neurological examination (Amiel-Tison, 1977, 2002). Only babies with a normal neurological examination were then recruited for the study. The study had full approval from the Clinical Ethics Committee of the Cork Teaching Hospitals and written informed parental consent was obtained for all infants studied Method of EEG recording Continuous video-eeg data was recorded using the NicoletOne EEG system from Cardinal Health. All the infants were in the supine position in their cots at the mother s bedside during each recording. All recordings commenced as soon as possible after birth and were continued for 1 2 hours to include AS and QS phases. EEG was recorded from seven scalp electrodes positioned using the system of electrode placement, modified for neonates (F4, F3, Cz, T4, T3, P4, P3). A reference electrode was placed between Cz and Fz. Silver silver chloride scalp electrodes were attached to the baby s scalp using a conductive water-soluble fixative paste and secured using an adhesive tape. A soft net was used to secure the electrodes if the baby was particularly active. Electrode impedance was maintained below 5 kohm. The data was sampled at 256 Hz and stored on the computer hard disk for off-line analysis. Two shoulder ECG electrodes were used to record a single channel heart rate signal. Respiratory movements were measured using a thoracic respiratory band. Simultaneous video was also recorded Visual EEG analysis The electrographic sleep states were categorised as follows: (1) active sleep (AS): combined theta and delta frequency waveforms with intermixed low-amplitude irregular segments (alpha frequency and faster rhythms); (2) quiet sleep (QS): combination of high amplitude slow waves, predominantly in the delta band, and tracé alternant segments (beta and theta activity alternating with delta activity); and (3) indeterminate sleep (IS) according to Scher et al. (2002). The EEG recordings were visually analysed and periods of AS, QS, IS and wakefulness were identified and annotated. The scoring criteria for EEG-sleep states, as defined by Scher et al. (1992), are summarised in Table 1. Four bipolar channels of EEG were analysed for symmetry, synchrony and continuity. The first hour of each recording was used to calculate the percentage of each sleep state present in each baby. The EEGs were divided into two groups: 1. Early : where EEGs were recorded <6 hours after birth. 2. Later : where EEGs were recorded between 6 and 12 hours after birth. All instances of artefact occurring during the entire recording were manually labelled and neglected from subsequent analysis. Figs. 1 and 2 give examples of AS and QS periods. Table 1 Scoring criteria for EEG-sleep states as defined by Scher et al. (1992). AS = active sleep; QS = quiet sleep; IS = indeterminate sleep. Sleep state Sleep state segment Onset of sleep state Active sleep Quiet sleep Complete sleep cycle Period from onset of AS to onset of QS, excluding any intervals Beginning of segment of AS = 3 consecutive minutes or 3 out of of IS at the transition 4 consecutive minutes scored as AS Period from the onset of QS to onset of AS, excluding any periods Beginning of segment of QS = 3 consecutive minutes or 3 out of of IS at the transition 4 consecutive minutes scored as QS Episode from onset of QS through a required period of AS (and IS if present) to onset of the next QS

3 1048 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) Fig. 1. Active sleep period. Ten minutes of artefact-free continuous EEG segments during AS and 10 min during QS were then selected from each infant s recording for frequency domain analysis which was performed using seven referential channels. These files were then exported to EDF format for off-line analysis Signal processing Each channel of EEG was band-pass filtered in the range Hz using a type II Chebyshev IIR filter. The EEG for each channel was then analysed in four-second non-overlapping epochs (N s = 4 s). The mean values for each EEG measure during selected Fig. 2. Quiet sleep period tracé alternant.

4 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) min of AS and 10 min of QS were calculated as the mean across all EEG channels and all 4s epochs for each 10 min AS or QS segment Quantitative EEG measures Ten-minute segments of EEG labelled as AS and QS were quantitatively analysed for each recording to assess the symmetry and synchrony of the EEG signal in both hemispheres. The same 10- min segments of AS and QS were also used to calculate a number of quantitative measures: SEF, H and d R. Several methods have been reported for quantitative analysis of the symmetry and synchrony of the EEG signal (Quian Quiroga et al., 2002; van Putten, 2007). In our study, we used the brain symmetry index (BSI) introduced by van Putten (2007) for analysis of symmetry and conventional cross-correlation (CC) (Northrop, 2003) for analysis of synchrony. Both measures were calculated on epochs of 4 s as defined in the previous section. BSI is defined as: BSI ¼ R L R þ L where R and L are the squared fast Fourier transform (FFT) coefficients of the signals from the right (F4-P4) and left (F3-P3) channels, respectively. If the signals in the considered channels are symmetrical, the index should be close to zero, while indices close to 1 or 1 would indicate left or right-shifted asymmetry. During CC analysis we were looking for the temporal lag d that maximizes the normalized CC coefficients: ^d ¼ arg maxðcc xy ðdþþ d P iððxðiþ mxþðyði dþ myþþ ¼ arg max qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d Pi ðxðiþ mxþ2 ðyði dþ myþ2 where x and y are two signals with corresponding mean values mx and my from the right and left channels, respectively. The lag is 0 when the signals are completely synchronous, while negative/positive lags indicate that there is a negative/positive delay between right and left channel signals. The higher the absolute value of the lag the larger the delay of one signal with respect to the other. The total EEG power in the range Hz was calculated for each 4 s EEG epoch. The SEF was then calculated as that frequency (in Hertz) below which 90% of the total spectral power resides. The relative power in the delta frequency band (d R ) was then calculated as the power in the delta frequency band (0.5 4 Hz) divided by the total spectral power in the range Hz, i.e. it is the proportion of the total power contained in the delta band (Hudson et al., 1983; Fell et al., 1996; Inder et al., 2003). The normalized spectral density estimate P(X) was used as a probability density function in the Shannon s entropy formula thus defining the spectral entropy (Eq. (1)) (Shannon, 1946; D Alessandro et al., 2003; Greene et al., 2008), normalized to the range [0 1]. HðXÞ ¼ 1 lnðn f Þ X PðXÞ lnðpðxþþ f where P(X) is an estimate of the PDF and is calculated by normalizing the power spectral density (PSD) estimate with respect to the total spectral power in the range Hz. A periodogram estimate of the PSD is calculated for each four-second epoch using a 1024-point FFT. N f is the number of frequency components in the PSD estimate. Pi ð1þ 2.6. Statistical analysis Mean values, standard deviation, minimum and maximum values were used to describe symmetrical data. Median, interquartile range, minimum and maximum values were reported on skewed data. Differences for each spectral measure under AS and QS conditions across all 30 infants were compared to 0 using a paired t-test. The assumptions underlying the test were checked for normality and validated. When the assumptions of normality were invalid (i.e. comparing the <6 hours after birth group to the 6-to-12 hours after birth group under AS and QS conditions), non-parametric Wilcoxon rank sum test was applied. 3. Results 3.1. Overall results EEGs for 30 neonates were analysed. Demographic characteristics for all infants are presented in Table 2. All recordings contained continuous, symmetrical and synchronous EEG activity at amplitudes ranging from 15 to 190 lv across all sleep states. Figs. 3 and 4 demonstrate the histograms of the BSI calculated across all recordings for AS and QS, respectively. In both AS and QS, the mean values were close to 0. The mean, minimum and maximum BSI values are presented in Table 3. The 2.5 and 97.5 percentiles [ 0.23; 0.22] for AS and [ 0.37; 0.31] for QS demonstrate that 95% of the data can be considered symmetrical in both AS and QS. The histograms of the cross-correlation lags calculated in samples across all EEGs for AS and QS are demonstrated in Figs. 5 and 6. Again, the signal can be considered synchronous as the most of the maximum cross-correlation coefficients appear at the lag equal to 0. Additionally, the means of the maximum cross-correlation values at the lag 0 (0.74 for AS and 0.7 for QS) indicate high-level synchrony of the EEG signals in both hemispheres. Apart from the symmetry and synchrony of the EEG signal, well-developed SWC was documented in all 30 recordings. The proportions of sleep states within 1 hour of recording are summarised in Table 4. There were no significant differences in the proportions of sleep states when comparing the early (<6 hours after birth) and the later (6-to-12 hours after birth) groups (Fig. 7). The results of Wilcoxon rank sum test comparing the proportions of sleep states between two groups are demonstrated in Table 5. Table 6 summarises the spectral measures data in all 30 infants. The mean SEF was significantly higher in AS than in QS (paired t- test, p < , 95% confidence interval for the difference Hz) and the irregularity and complexity of the EEG signal measured by H was greater in AS compared to QS (paired t-test, p < , 95% confidence interval for the difference ), indicating a higher level of disorganisation of the EEG signal during AS than during QS. Delta activity dominated in the neonatal EEG with less delta power during AS compared to QS (paired t-test, p < , 95% confidence interval for the difference 0.08 to 0.05). There were no significant statistical differences in SEF, H and d R when comparing the early (<6 hours after birth) group to the later (6-to-12 hours after birth) group under AS and QS conditions (Table 7). However, there were statistically significant differences in each of the parameters when comparing AS and QS (Figs. 8 10).

5 1050 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) Table 2 Demographic characteristics of the 30 normal neonates. $ = female; # = male; MOD = mode of delivery; SVD = spontaneous vaginal delivery; El CS = elective caesarean section; Em CS = emergency caesarean section; vacuum = vacuum delivery. Infant ID 4. Discussion Age (h) Gender Gestation Weight (g) MOD 1 5 M SVD 2 5 M El CS 3 4 M El CS 4 5 F SVD 5 5 M Vacuum 6 4 M El CS 7 5 F Em CS 8 3 F SVD 9 5 M El CS 10 4 M El CS Mean for <6 hours group 4.5 3$/7# F Em CS 12 9 M SVD 13 6 F SVD 14 9 M SVD 15 8 F SVD F SVD 17 7 M El CS 18 6 M SVD 19 7 M SVD F SVD 21 6 M El CS F SVD F SVD 24 9 M El CS M SVD M SVD F SVD 28 8 M SVD 29 7 F SVD M SVD Mean for 6 12 hours group 8.6 9$/11# Mean for all 30 infants $/18# This study has shown that normal neonatal EEG contains continuous symmetrical and synchronous activity with well-developed sleep wake cycling (SWC) as early as 6 hours after birth. We have also demonstrated that quantitative EEG analysis can easily distinguish between AS and QS in the healthy newborn. We Histogram of the brain symmetry indices for AS Histogram of the brain symmetry indices for QS Brain Symmetry Index Fig. 4. Histogram of the brain symmetry indices of the EEG signals calculated from the right (F4-P4) and left (F3-P3) channels across all recordings during QS. have seen no differences in either sleep state or quantitative measures when comparing the early (<6 hours) group to the later (6 12 hours) group. Our findings suggest that normal full-term neonates should have clearly defined SWC in the immediate postnatal period. In our population of infants, AS accounted for approximately half, and QS for approximately one third, of each one-hour recording, similar to the proportions reported by Scher et al. (2002). Visual analysis of the sleep neonatal EEG requires the expertise of a specialist unlike spectral analysis which can be an automated process. In healthy newborn babies SEF has been shown to change with behavioural state (Bell et al., 1991). However, this study included a mixed cohort of preterm and term babies with a small number of term neonates (n = 17); the EEGs were recorded on the third day after birth, and only eight four-second epochs were analysed. We have demonstrated clear differences in spectral measures under AS and QS conditions in the early postnatal period. This has implications for any further EEG studies that use quantitative analysis to assess EEG activity in the newborn. Several limitations of this study are recognised. We analysed only 1 hour of sleep from every baby. Longer EEG monitoring and serial studies within the early postnatal period might be useful in the assessment of the evolution of SWC early in life; however, in reality it is difficult to monitor healthy newborn babies for long periods of time in the immediate postnatal period. The number of neonates in both age groups was small. Nevertheless, appropriately applied statistical analysis revealed significant differences in all three spectral measures between AS and QS. Bigger sample groups could allow researchers to investigate further the influence of different modes of delivery and maternal anaesthesia during labour on neonatal SWC development. It may be argued that it would have been preferable to perform the sleep EEG study in the controlled environment of the sleep laboratory where it would be possible to control noise levels and Brain Symmetry Index Fig. 3. Histogram of the brain symmetry indices of the EEG signals calculated from the right (F4-P4) and left (F3-P3) channels across all recordings during AS. Table 3 Summary of the mean BSI values calculated across all infants under AS and QS conditions. SD = standard deviation; Min = minimum BSI value; Max = maximum BSI value. BSI across all infants Mean (SD) Min Max AS (0.12) QS (0.17)

6 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) Histogram of the cross-correlation lags for AS Cross-correlation lags (in samples) Fig. 5. Histogram of cross-correlation lags of all EEGs under AS condition. Histogram of the cross-correlation lags for QS Cross-correlation lags (in samples) Fig. 6. Histogram of cross-correlation lags of all EEGs under QS condition. 0.7 Fig. 7. The proportion of sleep states within 1 hour of recording for <6 and 6 12 hours groups. other factors. However, we deliberately chose not to separate the babies from their mothers in order to provide as little disruption as possible to the bonding process in the very early newborn period. We felt that it was essential to allow the babies to remain close to their mothers. We also demonstrated that it was possible to collect valid sleep-eeg data in this environment. To our knowledge, this study is the first to describe normative EEG data in healthy full-term babies so soon after birth. These findings are clinically important. We know that neuroprotective therapies must be administered to newborn babies with HIE within the first few hours of birth and that the EEG can identify those babies that are most suitable for treatment. It is important to describe the normative ranges in EEG activity in this age group in order to accurately identify those babies with HIE that will benefit most from this and future neuroprotective therapies. Table 4 Summary of the sleep-state proportions in normal neonates within 12 hours of birth. Q1 = 25th percentile; Q3 = 75th percentile; Min = minimum percentage; Max = maximum percentage. Median percentage (Q1, Q3) Min (%) Max (%) All 30 infants AS 48.5 (42.6, 63.5) QS 36.6 (28.1, 46.1) IS 2.1 ( ) Awake 4.7 ( ) <6 hours after birth (n = 10) AS 46.9 (41.3, 65.7) QS 35.3 (21.0, 46.9) IS 2.7 (0.0, 7.1) Awake 3.9 (0.0, 12.3) to-12 hours after birth (n = 20) AS 48.5 (44.6, 63.2) QS 36.9 (29.4, 45.8) IS 2.1 (0.0, 8.5) Awake 4.7 (1.0, 7.4) Table 5 Comparison of the proportions of EEG states between early (<6 hours after birth) and later (6-to-12 hours after birth) groups of normal neonates. Sleep state z-score q AS QS IS Awake Table 6 Spectral measure findings for AS and QS conditions in 30 normal neonates within 12 hours of birth. SEF = spectral edge frequency; H = spectral entropy; d R = relative delta power; Hz = Hertz; SD = standard deviation; Min = minimum values; Max = maximum values. Spectral measure AS QS Mean (SD) Min Max Mean (SD) Min Max SEF (Hz) 6.84 (1.02) (0.86) H 0.58 (0.03) (0.03) d R 0.73 (0.05) (0.04)

7 1052 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) Table 7 Spectral measures for early (<6 hours after birth) and later (6-to-12 hours after birth) groups of normal neonates. SEF = spectral edge frequency; H = spectral entropy; d R = relative delta power; Hz = Hertz; AS = active sleep; QS = quiet sleep; Q1 = 25th percentile; Q3 = 75th percentile; * = Wilcoxon rank sum test. Spectral measure AS QS <6 hours 6 12 hours Significance test * <6 hours 6 12 hours Significance test * z-score q z-score q Median SEF (Hz) (Q1, Q3) 6.89 (6.0, 7.68) 6.58 (6.09, 7.21) (5.17, 5.56) 5.23 (4.99, 6.12) Median H (Q1, Q3) 0.60 (0.56, 0.61) 0.58 (0.55, 0.60) (0.54, 0.55) 0.54 (0.52, 0.56) Median d R (Q1, Q3) 0.71 (0.69, 0.75) 0.74 (0.70, 0.78) (0.77, 0.80) 0.81 (0.78, 0.82) Fig. 8. Box plot of median SEF values for all infants in the early (<6 hours) and the later (6 12 hours) groups under AS and QS conditions. Fig. 10. Box plot of median relative delta power values for all infants in the early (<6 hours) and the later (6 12 hours) groups under AS and QS conditions. 5. Conclusion The EEG of the healthy full-term newborn, recorded within the 12 hours of birth, is continuous, symmetrical and bilaterally synchronous with amplitudes ranging from 15 to 190 lv across all sleep states. The early neonatal EEG contains well-developed sleep/wake architecture as early as 6 hours after birth. There are significant differences between active sleep and quiet sleep in several quantitative EEG measures including spectral edge frequency, spectral entropy and relative delta power. Acknowledgement This study was supported by grants from Science Foundation Ireland (SFI/05/PICA/1836) and Health Research Board, Ireland (CRT/2008/31). References Fig. 9. Box plot of median H values for all infants in the early (<6 hours) and the later (6 12 hours) groups under AS and QS conditions. Amiel-Tison C. Neurologic examination of the newborn infant. Rev Prat 1977;27(33): Amiel-Tison C. Update of the Amiel-Tison neurologic assessment for the term neonate or at 40 weeks corrected age. Pediatr Neurol 2002;27(3): Armitage R, Roffwarg H. Distribution of period-analysed delta activity during sleep. Sleep 1993;16: Bell A, McClure B, McCullagh P, McClelland R. Spectral edge frequency of the EEG in healthy neonates and variation with behavioural state. Biol Neonate 1991;60:69 74.

8 I. Korotchikova et al. / Clinical Neurophysiology 120 (2009) D Alessandro M, Esteller R, Vachtsevanos G, Hinson A, Echauz J, Litt B. Epileptic seizure prediction using hybrid feature selection over multiple intracranial EEG electrode contacts: a report of four patients. IEEE Trans Biomed Eng 2003: Fell J, Roschke J, Mann K, Schaffner C. Discrimination of sleep stages: a comparison between spectral and nonlinear EEG measures. Electroencephalogr Clin Neurophysiol 1996;98: Gluckman P, Wyatt J, Azzopardi D, Ballard R, Edwards A, Ferriero D, et al. Selective head cooling with mild systemic hypothermia after neonatal encephalopathy. Lancet 2005;365: Greene B, Faul S, Marnane L, Lightbody G, Korotchikova I, Boylan G. A comparison of quantitative EEG features for neonatal seizure detection. Clin Neurophysiol 2008;119(6): Gurman G. Assessment of depth of general anesthesia observations on processed EEG and spectral edge frequency. Int J Clin Monitor Comput 1994;11: Hudson R, Stanski D, Saidman L, Meathe E. A model for studying depth of anesthesia and acute tolerance to thiopental. Anesthesiology 1983;59: Inder T, Buckland L, Williams C, Spencer C, Gunning M, Darlow B, et al. Lowered electroencephalographic spectral edge frequency predicts the presence of cerebral white matter injury in premature infants. Pediatrics 2003;111: Lombroso C. Neonatal polygraphy in full-term and preterm infants: a review of normal and abnormal findings. J Clin Neurophysiol 1985;2: Long C, Shah N, Loughlin C, Spydell J, Bedford R. A comparison of EEG determinants of near-awakening from isoflurane and fentanyl anesthesia spectral edge, median power frequency, and delta-ratio. Anesth Analg 1989;69: Menache C, Bourgeois B, Volpe J. Prognostic value of neonatal discontinuous EEG. Pediatr Neurol 2002;27: Northrop R. Signals and systems analysis in biomedical engineering. Boca Raton, FL: CRC Press; Osredkar D, Toet M, van Rooij L, van Huffelen A, Groenendaal F, de Vries L. Sleep wake cycling on amplitude-integrated electroencephalography in term newborns with hypoxic-ischaemic encephalopathy. Pediatrics 2005;115: Pezzani C, Radvanyi-Bouvet M, Relier J, Monod N. Neonatal electroencephalography during the first twenty-four hours of life in full-term newborn infants. Neuropediatrics 1986;17:11 8. Pressler R, Boylan G, Morton M, Binnie C, Rennie J. Early serial EEG in hypoxic ischaemic encephalopathy. Clin Neurophysiol 2001;112:31 7. Quian Quiroga R, Kreuz T, Grassberger P. Event synchronization: a simple and fast method to measure synchronicity and time delay patterns. Phys Rev E 2002;66: Scher M, Steppe D, Dahl R, Asthana S, Guthrie R. Comparison of EEG sleep measures in healthy full-term and preterm infants at matched conceptional ages. Sleep 1992;15: Scher M. Neonatal encephalopathies as classified by EEG-sleep criteria: severity and timing based on clinical/pathologic correlations. Pediatr Neurol 1994;11: Scher M. Normal electrographic polysomnographic patterns in preterm and fullterm infants. Semin Pediatr Neurol 1996;3:2 12. Scher M, Steppe D, Beggarly M, Salerno D, Banks D. Neonatal EEG-sleep disruption mimicking hypoxic-ischemic encephalopathy after intrapartum asphyxia. Sleep Med 2002;3: Shannon C. A mathematical theory of communication. Bell Syst Tech J 1946: ter Horst H, Sommer C, Bergman K, Fock J, van Weerden T, Bos A. Prognostic significance of amplitude-integrated EEG during the first 72 hours after birth in severely asphyxiated neonates. Pediatr Res 2004;55: Thordstein M, Flisberg A, Lofgren N, Bagenholm R, Lindecrantz K, Wallin B, et al. Spectral analysis of burst periods in EEG from healthy and post-asphyctic fullterm neonates. Clin Neurophysiol 2004;115: Thorngren-Jerneck K, Hellstrom-Westas L, Ryding E, Rosen I. Cerebral glucose metabolism and early EEG/aEEG in term newborn infants with hypoxicischemic encephalopathy. Pediatr Res 2003;54: van Putten M. The revised brain symmetry index. Clin Neurophysiol 2007;118: Varner J, Ellingson R, Dahahy T, Nelson B. Interhemispheric amplitude symmetry in the EEGs of full-term newborns. Electroencephalogr Clin Neurophysiol 1977;43: Varner J, Peters J, Ellingson R. Interhemispheric synchrony in the EEGs of full-term newborns. Electroencephalogr Clin Neurophysiol 1978;45: Watanabe K, Miyazaki S, Hara K, Hakamada S. Behavioral state cycles, background EEGs and prognosis of newborns with perinatal hypoxia. Electroencephalogr Clin Neurophysiol 1980;49: Wertheim D, Mercuri E, Faundez J, Rutherford M, Acolet D, Dubowitz L. Prognostic value of continuous electroencephalographic recording in full-term infants with hypoxic-ischemic encephalopathy. Arch Dis Child 1994;71:F97 F102. Wyatt J, Gluckman P, Liu P, Azzopardi D, Ballard R, Edwards A, et al. Determination of outcomes after head cooling for neonatal encephalopathy. Pediatrics 2007;119: