An Automated Quiet Sleep Detection Approach in Preterm Infants as a Gateway to Assess Brain Maturation

Anneleen Dereymaeker, Kirubin Pillay, Jan Vervisch, Sabine Van Huffel, Gunnar Naulaers, Katrien Jansen, Maarten De Vos, Anneleen Dereymaeker, Kirubin Pillay, Jan Vervisch, Sabine Van Huffel, Gunnar Naulaers, Katrien Jansen, Maarten De Vos

Abstract

Sleep state development in preterm neonates can provide crucial information regarding functional brain maturation and give insight into neurological well being. However, visual labeling of sleep stages from EEG requires expertise and is very time consuming, prompting the need for an automated procedure. We present a robust method for automated detection of preterm sleep from EEG, over a wide postmenstrual age ([Formula: see text] age) range, focusing first on Quiet Sleep (QS) as an initial marker for sleep assessment. Our algorithm, CLuster-based Adaptive Sleep Staging (CLASS), detects QS if it remains relatively more discontinuous than non-QS over PMA. CLASS was optimized on a training set of 34 recordings aged 27-42 weeks PMA, and performance then assessed on a distinct test set of 55 recordings of the same age range. Results were compared to visual QS labeling from two independent raters (with inter-rater agreement [Formula: see text]), using Sensitivity, Specificity, Detection Factor ([Formula: see text] of visual QS periods correctly detected by CLASS) and Misclassification Factor ([Formula: see text] of CLASS-detected QS periods that are misclassified). CLASS performance proved optimal across recordings at 31-38 weeks (median [Formula: see text], median MF 0-0.25, median Sensitivity 0.93-1.0, and median Specificity 0.80-0.91 across this age range), with minimal misclassifications at 35-36 weeks (median [Formula: see text]). To illustrate the potential of CLASS in facilitating clinical research, normal maturational trends over PMA were derived from CLASS-estimated QS periods, visual QS estimates, and nonstate specific periods (containing QS and non-QS) in the EEG recording. CLASS QS trends agreed with those from visual QS, with both showing stronger correlations than nonstate specific trends. This highlights the benefit of automated QS detection for exploring brain maturation.

Keywords: CLASS; EEG; automated sleep detection; brain maturation; preterm neonate; quiet sleep.

Figures

Fig. 1
Fig. 1
Histogram of the total number of EEG recordings used in the study, ordered by PMA. There are a total of 89 recordings ranging from 27 to 42 weeks PMA.
Fig. 2
Fig. 2
(Color online) (a) Flowchart of the stages of EEG processing by CLASS. (b) Illustration of the Adaptive Segmentation (ASG) stage for a 100 s period of EEG in a single channel. Red line denote the ASG segment boundaries. (c) Illustration of a Cluster-Time Profile for a 2 h epoch of EEG from a single channel. Features are extracted from each segment defined by ASG and then clustered and the corresponding segment cluster labels are then plotted over time for each sample. (d) The average cluster-time profile determined by taking the mean profile across all channels. Regions of increasing cluster fluctuation (shaded) correspond to higher EEG discontinuity and QS periods. (e) De-trended signal after subtraction of the average channel from its running mean. (f) The square of the zeroed signal with the signal envelope shown by a red curve. (g) The signal envelope of a complete 4 h EEG recording, with the mean threshold to estimate the QS periods shown in red. Here, the 4 h signal envelope is formed by stitching the signal envelope processed for every 2 h epoch of EEG. The first 2 h of the stitched envelope shown in this figure correspond to the envelope derived in (f). (h) The QS periods as estimated by CLASS after thresholding with the mean of the signal envelope. Estimated QS periods are shaded. (i) The shaded QS periods as visually estimated by the clinician using the full PSG recording.
Fig. 3
Fig. 3
Illustration of the ASR method. (a) Top: A 30-min epoch of bandpass filtered (1–40 Hz) EEG in a single channel, before ASR is applied. High power artifacts are shaded. Bottom: The bandpass filtered signal after ASR is applied. The same shaded artifacts are now reduced while surrounding clean periods of the signal remain intact. (b) Illustration of the cleaning procedure of ASR on the EEG recording. Reconstruction metrics are calculated within the sliding window S in order to clean the sample of data along the dotted line denoted by s. As the sliding window moves sample-by-sample across the recording, the metrics are updated and the new sample s is cleaned.
Fig. 4
Fig. 4
Assessing the performance of CLASS on a test set of 55 recordings aged 27–42 weeks PMA. (a) ROC of CLASS performance by varying the detection threshold while keeping all other optimized parameters constant. ROC curves for each recording in the test set (in gray) are shown, and resulting median ROC curve (in black). The AUC of the median ROC curve is also presented. (b) CLASS performance with respect to PMA denoting Sensitivity (Sens), Specificity (Spec), DF and MF. DF and MF denote Detection Factor and Misclassification Factor measures, respectively. DF measures the proportion of visually labeled QS periods correctly detected by CLASS, while MF measures the proportion of CLASS-detected periods that do not correspond to the visual QS periods (i.e. are misclassifications). Error bars denote the medians and IQRs.

References

    1. Aarnoudse-Moens CSH, Weisglas-Kuperus N, Van Goudoever JB, Oosterlaan J. Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics. 2009;124(2):717–728. doi: 10.1542/peds.2008-2816.
    1. Back SA, Miller SP. Brain injury in premature neonates: A primary cerebral dysmaturation disorder? Ann Neurol. 2014;75(4):469–486. doi: 10.1002/ana.24132.
    1. Larroque B, Ancel PY, Marret S, et al. Neurodevelopmental disabilities and special care of 5-year-old children born before 33 weeks of gestation (the EPIPAGE study): A longitudinal cohort study. Lancet. 2008;371(9615):813–820. doi: 10.1016/S01406736(08)60380-3.
    1. de Kieviet JF, Zoetebier L, van Elburg RM, Vermeulen RJ, Oosterlaan J. Brain development of very preterm and very low-birthweight children in childhood and adolescence: A meta-analysis. Dev Med Child Neurol. 2012;54(4):313–323. doi: 10.1111/j.14698749.2011.04216.x.
    1. Hellström-Westas L, Rosén I. Electroencephalography and brain damage in preterm infants. Early Hum Dev. 2005;81(3):255–261. doi: 10.1016/j.earlhumdev.2005.01.006.
    1. Malk K, Metsäranta M, Vanhatalo S. Drug effects on endogenous brain activity in preterm babies. Brain Dev. 2014;36(2):116–123. doi: 10.1016/j.braindev.2013.01.009.
    1. Shany E, Meledin I, Gilat S, Yogev H, Golan A, Berger I. In and ex utero maturation of premature infants electroencephalographic indices. Clin Neurophysiol. 2014;125(2):270–276. doi: 10.1016/j.clinph.2013.06.185.
    1. Scher MS. Ontogeny of EEG-sleep from neonatal through infancy periods. Sleep Med. 2008;9(6):615–636. doi: 10.1016/j.sleep.2007.08.014.
    1. Shellhaas RA, Burns JW, Barks JDE, Chervin RD. Quantitative sleep stage analyses as a window to neonatal neurologic function. Neurology. 2014;82(5):390–395. doi: 10.1212/WNL.0000000000000085.
    1. Hoppenbrouwers T, Hodgman JE, Rybine D, et al. Sleep architecture in term and preterm infants beyond the neonatal period: The influence of gestational age, steroids, and ventilatory support. Sleep. 2005;28(11):1428–1436.
    1. Graven SN, Browne JV. Sleep and brain development, The critical role of sleep in fetal and early neonatal brain development. Newborn Infant Nurs Rev. 2008;8(4):173–179. doi: 10.1053/j.nainr.2008.10.008.
    1. Grigg-Damberger MM. The visual scoring of sleep in infants 0 to 2 months of age. J Clin Sleep Med. 2016;12(3):429–445. doi: 10.5664/jcsm.5600.
    1. Kostovic I, Judas M. The development of the subplate and thalamocortical connections inthe human foetal brain. Acta Paediatr Int J Paediatr. 2010;99(8):1119–1127. doi: 10.1111/j.1651-2227.2010.01811.x.
    1. Koolen N, Dereymaeker A, Räsänen O, et al. Interhemispheric synchrony in the neonatal EEG revisited: Activation synchrony index as a promising classifier. Front Hum Neurosci. 2014 Dec;8:1030. doi: 10.3389/fnhum.2014.01030.
    1. Holmes G, Lombroso C. Prognostic value of background patterns in the neonatal EEG. J Clin Neurophysiol. 1993;10:323–352.
    1. Hayashi-Kurahashi N, Kidokoro H, Kubota T, et al. EEG for predicting early neurodevelopment in preterm infants: An observational cohort study. Pediatrics. 2012;130(4):e891–e897. doi: 10.1542/peds.2012-1115.
    1. O’Toole JM, Boylan GB, Vanhatalo S, Stevenson NJ. Estimating functional brain maturity in very and extremely preterm neonates using automated analysis of the electroencephalogram. Clin Neurophysiol. 2016;127(8):2910–2918. doi: 10.1016/j.clinph.2016.02.024.
    1. Dereymaeker A, Koolen N, Jansen K, et al. The suppression curve as a quantitative approach- for measuring brain maturation in preterm infants. Clin Neurophysiol. 2016;127(8):2760–2765. doi: 10.1016/j.clinph.2016.05.362.
    1. Watanabe K. Neurophysiological aspects of neonatal seizures. Brain Dev. 2014;36(5):363–371. doi: 10.1016/j.braindev.2014.01.016.
    1. Kidokoro H, Inder T, Okumura A, Watanabe K. What does cyclicity on amplitude-integrated EEG mean? J Perinatol. 2012;32(8):565–569. doi: 10.1038/jp.2012.25.
    1. Scher MS. Neurophysiological assessment of brain function and maturation: I. A measure of brain adaptation in high risk infants. Pediatr Neurol. 1997;16(3):191–198. doi: 10.1016/S0887-8994(97)00008-8.
    1. Palmu K, Kirjavainen T, Stjerna S, Salokivi T, Vanhatalo S. Sleep wake cycling in early preterm infants: Comparison of polysomnographic recordings with a novel EEG-based index. Clin Neurophysiol. 2013;124(9):1807–1814. doi: 10.1016/j.clinph.2013.03.010.
    1. Turnbull J, Loparo K, Johnson M, Scher M. Automated detection of tracé alternant during sleep in healthy full-term neonates using discrete wavelet transform. Clin Neurophysiol. 2001;112(10):1893–1900. doi: 10.1016/S1388-2457(01)00641-1.
    1. Palmu K, Stevenson N, Wikström S, Hellström-Westas L, Vanhatalo S, Palva JM. Optimization of an NLEO-based algorithm for automated detection of spontaneous activity transients in early preterm EEG. Physiol Meas. 2010;31(11):85–93. doi: 10.1088/0967-3334/31/11/N02.
    1. Krajca V, Petranek S, Paul K, Matousek M, Mohylova J, Lhotska L. Automatic detection of sleep stages in neonatal EEG using the structural time profiles. 2005 IEEE Engineering in Medicine and Biology 27th Annual Conf; 2005. pp. 6014–6016.
    1. Krajca V, Petranek S, Patakova I, Varri A. Automatic identification of significant graphoelements in multichannel EEG recordings by adaptive segmentation and fuzzy clustering. Int J Biomed Comput. 1991;28(1–2):71–89. doi: 10.1016/0020-7101(91)90028-D.
    1. Krajca V, Petránek S, Mohylová J, Paul K, Gerla V, Lhotská L. Neonatal EEG sleep stages modelling by temporal profiles. Computer Aided Systems Theory — EUROCAST; Berlin, Heidelberg: Springer; 2007. pp. 195–201. 2007.
    1. André M, Lamblin M-D, D’Allest AM, et al. Electroencephalography in premature and full-term infants. Developmental features and glossary. Clin Neurophysiol. 2010;40(2):59–124. doi: 10.1016/j.neucli.2010.02.002.
    1. Werth J, Atallah L, Andriessen P, Long X, Zwartkruis-pelgrim E, Aarts RM. Unobtrusive sleep state measurements in preterm infants — A review. Sleep Med Rev. 2016 doi: 10.1016/j.smrv.2016.03.005.
    1. Ludington-Hoe SM, Johnson M, Morgan K, et al. Neurophysiologic assessment of neonatal sleep organization: Preliminary results of a randomized, controlled trial of skin contact withpreterm infants. Pediatrics. 2006;117(5):e909–e923. doi: 10.1542/peds.2004-1422.
    1. Scher MS, Steppe DA, Banks DL, Guthrie RD, Sclabassi RJ. Maturational trends of EEG-sleep measures in the healthy preterm neonate. Pediatr Neurol. 1995;12(4):314–322. doi: 10.1016/0887-8994(95)00052-H.
    1. Scher MS, Steppe DA, Banks DL. Prediction of lower developmental performances of healthy-neonates by neonatal EEG-sleep measures. Pediatr Neurol. 1996;14(2):137–144. doi: 10.1016/0887-8994(96)00013-6.
    1. Korotchikova I, Stevenson NJ, Livingstone V, Ryan CA, Boylan GB. Sleep–wake cycle of the healthy term newborn infant in the immediate postnatal period. Clin Neurophysiol. 2015 doi: 10.1016/j.clinph.2015.12.015.
    1. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 2008;33(1):159–174. doi: 10.2307/2529310.
    1. De Vos M, Deburchgraeve W, Cherian PJ, et al. Automated artifact removal as preprocessing refines neonatal seizure detection. Clin Neurophysiol. 2011;122(12):2345–2354. doi: 10.1016/j.clinph.2011.04.026.
    1. Lagerlund TD, Sharbrough FW, Busacker NE. Spatial filtering of multichannel electroencephalographic recordings through principal component analysis by singular value decomposition. J Clin Neurophysiol. 1997;14(1):73–82.
    1. Kothe CA, Makeig S. BCILAB: A platform for brain-computer interface development. J Neural Eng. 2013;10(5):56014. doi: 10.1088/1741-2560/10/5/056014.
    1. Mullen T, Kothe C, Chi YM, et al. Real-time modeling and 3D visualization of source dynamics and connectivity using wearable EEG. Annual Int Conf IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conf. 2013:2184–2187. doi: 10.1109/EMBC.2013.6609968.
    1. Matic V, Cherian PJ, Jansen K, et al. Improving reliability of monitoring background EEG dynamics in asphyxiated infants. IEEE Trans Biomed Eng. 2016;63(5):973–983.
    1. Scher MS, Dokianakis SG, Sun M, Steppe DA, Guthrie RD, Sclabassi RJ. Computer classification of sleep in preterm and full-term neonates at similar postconceptional term ages. Sleep. 1996;19(1):18–25.
    1. Husain AM. Review of neonatal EEG. Am J Electroneurodiagnostic Technol. 2005;45(1):12–35. doi: 10.1080/1086508X.2005.11079505.
    1. Rangayyan RM. Biomedical Signal Analysis. 2nd edn. JohnWiley & Sons inc.; Hoboken, New Jersey: 2015.
    1. Jennekens W, Niemarkt HJ, Engels M, Pasman JW, van Pul C, Andriessen P. Topography of maturational changes in EEG burst spectral power of the preterm infant with a normal follow-up at 2 years of age. Clin Neurophysiol. 2012;123(11):2130–2138. doi: 10.1016/j.clinph.2012.03.018.
    1. Koolen N, Jansen K, Vervisch J, et al. Line length as a robust method to detect high-activity events: Automated burst detection in premature EEG recordings. Clin Neurophysiol. 2014;125(10):1985–1994. doi: 10.1016/j.clinph.2014.02.015.
    1. Stevenson NJ, Palmu K, Wikström S, Hellström-Westas L, Vanhatalo S. Measuring brain activity cycling (BAC) in long term EEG monitoring of preterm babies. Physiol Meas. 2014;35(7):1493–1508. doi: 10.1088/0967-3334/35/7/1493.
    1. Gerla V, Bursa M, Lhotska L, Paul K, Krajca V. Newborn sleep stage classification using hybrid evolutionary approach. Int J Bioelectromagn. 2007;9(1):25–26.
    1. Löfhede J, Thordstein M, Löfgren N, et al. Automatic classification of background EEG activity in healthy and sick neonates. J Neural Eng. 2010;7(1):16007. doi: 10.1088/1741-2560/7/1/016007.
    1. Piryatinska A, Terdik G, Woyczynski WA, Loparo KA, Scher MS, Zlotnik A. Automated detection of neonate EEG sleep stages. Comput Methods Prog Biomed. 2009;95(1):31–46. doi: 10.1016/j.cmpb.2009.01.006.
    1. Scher MS. Understanding sleep ontogeny to assess brain dysfunction in neonates and infants. J Child Neurol. 1998;13:467–474. doi: 10.1177/088307389801301001.
    1. dos Santos AÁ, Khan RL, Rocha G, Nunes ML. Behavior and EEG concordance of active and quiet sleep in preterm very low birth weight and full-term neonates at matched conceptional age. Early Hum Dev. 2014;90(9):507–510. doi: 10.1016/j.earlhumdev.2014.06.014.
    1. Curzi-Dascalova L, Figueroa JM, Eiselt M, et al. Sleep state organization in premature infants of less than 35 weeks’ gestational age. Pediatr Res. 1993;34(5):624–628. doi: 10.1203/00006450-199311000-00013.
    1. Tolonen M, Palva JM, Andersson S, Vanhatalo S. Development of the spontaneous activity transients and ongoing cortical activity in human preterm babies. Neuroscience. 2007;145(3):997–1006. doi: 10.1016/j.neuroscience.2006.12.070.

Source: PubMed

Sponsorer och medarbetare

Medicinska tillstånd

Läkemedelsinterventioner

3
Prenumerera