Search

INTRODUCTION: Existing automated seizure detection algorithms report sensitivities between 43% and 77% and specificities between 56% and 90%. The algorithms suffer from false alarms when applied to neonatal EEG because of the high degree of nurse handling and rhythmic patting used to soothe neonates. Computer vision technology that quantifies movement in real time could distinguish artifactual motion and improve automated neonatal seizure detection algorithms. METHODS: The authors used video EEG recordings from 43 neonates undergoing monitoring for seizures as part of the NEOLEV2 clinical trial. The Persyst neonatal automated seizure detection algorithm ran in real time during study EEG acquisitions. Computer vision algorithms were applied to extract detailed accounts of artifactual movement of the neonate or people near the neonate though dense optical flow estimation. RESULTS: Using the methods mentioned above, 197 periods of patting activity were identified and quantified, of which 45 generated false-positive automated seizure detection events. A binary patting detection algorithm was trained with a subset of 470 event videos. This supervised detection algorithm was applied to a testing subset of 187 event videos with 8 false-positive events, which resulted in a 24% reduction in false-positive automated seizure detections and a 50% reduction in false-positive events caused by neonatal care patting, while maintaining 11 of 12 true-positive seizure detection events. CONCLUSIONS: This work presents a novel approach to improving automated seizure detection algorithms used during neonatal video EEG monitoring. This artifact detection mechanism can improve the ability of a seizure detector algorithm to distinguish between artifact and true seizure activity.

ObjectiveEstablish if serial, multichannel video electroencephalography (EEG) in preterm infants can accurately predict 2-year neurodevelopmental outcome.Design and patientsEEGs were recorded at three time points over the neonatal course for infants SettingNeonatal intensive care unit at Cork University Maternity Hospital, Ireland.Main outcome measuresBayley Scales of Infant Development III at 2 years’ corrected age.ResultsSixty-seven infants were prospectively enrolled in the study and 57 had follow-up available (median GA 28.9 weeks (IQR 26.5–30.4)). Forty had normal outcome, 17 had abnormal outcome/died. All EEG time points were individually predictive of abnormal outcome; however, the 35-week EEG performed best. The area under the receiver operating characteristic curve (AUC) for this time point was 0.91 (95% CI 0.83 to 1), pConclusionMultichannel EEG is a strong predictor of 2-year outcome in preterm infants particularly when recorded around 35 weeks’ PMA. Infants at high risk of brain injury may benefit from early postnatal EEG recording which, if normal, is reassuring. Postnatal clinical complications can contribute to poor outcome; therefore, we state that a later EEG around 35 weeks has a role to play in prognostication.

Funding Information: Funding for this research was provided by the Finnish Cultural Foundation ( 00181077 ), Finnish Pediatric Foundation, the Finnish Academy ( 313242 , 288220 , 321235 ), Aivosäätiö, Neuroscience Center at University of Helsinki , Helsinki University Hospital, Helsinki University Hospital Research Funds, HUS and University of Helsinki Researcher position, Sigrid Juselius Foundation and European Union's Horizon 2020 Research and Innovation Programme ( H2020-MCSA-ITN-813483 and H2020-MCSA-IF-656131 ). The study sponsors had no involvement in the study design, nor in the collection, analysis or interpretation of data. Additionally we acknowledge the computational resources provided by the Aalto Science-IT project. Publisher Copyright: © 2022 The Authors Neonatal seizure detection algorithms (SDA) are approaching the benchmark of human expert annotation. Measures of algorithm generalizability and non-inferiority as well as measures of clinical efficacy are needed to assess the full scope of neonatal SDA performance. We validated our neonatal SDA on an independent data set of 28 neonates. Generalizability was tested by comparing the performance of the original training set (cross-validation) to its performance on the validation set. Non-inferiority was tested by assessing inter-observer agreement between combinations of SDA and two human expert annotations. Clinical efficacy was tested by comparing how the SDA and human experts quantified seizure burden and identified clinically significant periods of seizure activity in the EEG. Algorithm performance was consistent between training and validation sets with no significant worsening in AUC (p > 0.05, n = 28). SDA output was inferior to the annotation of the human expert, however, re-training with an increased diversity of data resulted in non-inferior performance (Δκ = 0.077, 95% CI: −0.002-0.232, n = 18). The SDA assessment of seizure burden had an accuracy ranging from 89 to 93%, and 87% for identifying periods of clinical interest. The proposed SDA is approaching human equivalence and provides a clinically relevant interpretation of the EEG.

Inter-areal synchronization by phase-phase correlations (PPCs) of cortical oscillations mediates many higher neurocognitive functions, which are often affected by prematurity, a globally prominent neurodevelopmental risk factor. Here, we used electroencephalography to examine brain-wide cortical PPC networks at term-equivalent age, comparing human infants after early prematurity to a cohort of healthy controls. We found that prematurity affected these networks in a sleep state-specific manner, and the differences between groups were also frequency-selective, involving brain-wide connections. The strength of synchronization in these networks was predictive of clinical outcomes in the preterm infants. These findings show that prematurity affects PPC networks in a clinically significant manner, suggesting early functional biomarkers of later neurodevelopmental compromise that may be used in clinical or translational studies after early neonatal adversity.

Electroencephalography (EEG) is the reference tool for the analysis of brain function, reflecting normal and pathological neuronal network activity. During the neonatal period, EEG patterns evolve weekly, according to gestational age. The first analytical criteria for the various maturational stages and standardized neonatal EEG terminology were published by a group of French neurophysiologists training in Paris (France) in 1999. These criteria, defined from analog EEG, were completed in 2010 with digital EEG analysis. Since then, this work has continued, aided by the technical progress in EEG acquisition, the improvement of knowledge on the maturating processes of neuronal networks, and the evolution of critical care. In this review, we present an exhaustive and didactic overview of EEG characteristics from extremely premature to full-term infants. This update is based on the scientific literature, enhanced by the study of normal EEGs of extremely premature infants by our group of neurophysiologists. For educational purposes, particular attention has been paid to illustrations using new digital tools.

Highlights • Higher power of cortical bursts after postnatal day 3 predicted worse cognitive, language and motor outcomes. • Association between cortical bursts and outcome was independent of structural MRI findings. • EEG may provide additional information by indexing persistent active mechanisms that support recovery or exacerbate damage., Objective Early prediction of neurological deficits following neonatal hypoxic-ischemic encephalopathy (HIE) may help to target support. Neonatal animal models suggest that recovery following hypoxia-ischemia depends upon cortical bursting. To test whether this holds in human neonates, we correlated the magnitude of cortical bursting during recovery (≥postnatal day 3) with neurodevelopmental outcomes. Methods We identified 41 surviving infants who received therapeutic hypothermia for HIE (classification at hospital discharge: 19 mild, 18 moderate, 4 severe) and had 9-channel electroencephalography (EEG) recordings as part of their routine care. We correlated burst power with Bayley-III cognitive, motor and language scores at median 24 months. To examine whether EEG offered additional prognostic information, we controlled for structural MRI findings. Results Higher power of central and occipital cortical bursts predicted worse cognitive and language outcomes, and higher power of central cortical bursts predicted worse motor outcome, all independently of structural MRI findings. Conclusions Clinical EEG after postnatal day 3 may provide additional prognostic information by indexing persistent active mechanisms that either support recovery or exacerbate brain damage, especially in infants with less severe encephalopathy. Significance These findings could allow for the effect of clinical interventions in the neonatal period to be studied instantaneously in the future.

Aim: Based on published reports and our own observations, we aimed to assign the graphical elements of EEG into normal and abnormal groups and then identify up to five main graphic elements in each group to facilitate visual analysis and interpretation of EEG in young children.Materials and methods. We searched for the relevant sources in the Medline and Medscape databases using the following keywords: «neonatal EEG», «neonatal seizures», «theta burst», «delta brushes», «trace discontinue», «burst-suppression», hypsarrhythmia», etc. Our own research was conducted using an Encephalan-EEGR-19/26 encephalograph equipped with children size gold cup electrodes with a diameter of 0.6 cm. Encephalograms were recorded from 10 electrodes according to the international “10-20” system.Results and discussion. In early childhood EEGs, two large groups of EEG graphic elements can be discerned: the likely normative graphic elements and patterns of pediatric EEG (normal patterns) and the likely non-normative (abnormal) graphic elements and patterns of pediatric EEG. In the likely normative group, the main features are represented by: theta bursts, delta brushes, the “intermittent EEG curve” pattern, the occipital theta rhythm, and slow-sleep waves. In the likely non-normative group, those are: paroxysmal EEG graphic elements, asynchronous EEG patterns, spike-wave discharges, 3 Hz peak waves, hypsarrhythmia, burst-suppression pattern, rolandic occipital spikes, and a slowing rhythm pattern.Conclusion. Along with the numerous attempts to characterize the age-dependent graphic elements at an early age, there are few reports concerning older children and adults. Here we try to overcome this discrepancy by identifying two large groups of graphic elements in EEG that are similar between infants and children of an older age. Such an approach may contribute to a better understanding of normal and pathological ontogenesis.

The conventional assessment of preterm somatosensory functions using averaged cortical responses to electrical stimulation ignores the characteristic components of preterm somatosensory evoked responses (SERs). Our study aimed to systematically evaluate the occurrence and development of SERs after tactile stimulus in preterm infants. We analysed SERs performed during 45 electroencephalograms (EEGs) from 29 infants at the mean post-menstrual age of 30.7 weeks. Altogether 2,087 SERs were identified visually at single trial level from unfiltered signals capturing also their slowest components. We observed salient SERs with a high amplitude slow component at a high success rate after hand (95%) and foot (83%) stimuli. There was a clear developmental change in both the slow wave and the higher frequency components of the SERs. Infants with intraventricular haemorrhage (IVH; eleven infants) had initially normal SERs, but those with bilateral IVH later showed a developmental decrease in the ipsilateral SER occurrence after 30 weeks of post-menstrual age. Our study shows that tactile stimulus applied at bedside elicits salient SERs with a large slow component and an overriding fast oscillation, which are specific to the preterm period. Prior experimental research indicates that such SERs allow studying both subplate and cortical functions. Our present findings further suggest that they might offer a window to the emergence of neurodevelopmental sequalae after major structural brain lesions and, hence, an additional tool for both research and clinical neurophysiological evaluation of infants before term age.

Objective To develop and validate an automated method for bedside monitoring of sleep state fluctuations in neonatal intensive care units. Methods A deep learning-based algorithm was designed and trained using 53 EEG recordings from a long-term (a)EEG monitoring in 30 near-term neonates. The results were validated using an external dataset from 30 polysomnography recordings. In addition to training and validating a single EEG channel quiet sleep, we constructed Sleep State Trend (SST), a bedside-ready means for visualising classifier outputs. Results The accuracy of quiet sleep detection in the training data was 90%, and the accuracy was comparable (85-86 %) in all bipolar derivations available from the 4-electrode recordings. The algorithm generalised well to an external dataset, showing 81% overall accuracy despite different signal derivations. SST allowed an intuitive, clear visualisation of the classifier output. Conclusions Fluctuations in sleep states can be detected at high fidelity from a single EEG channel, and the results can be visualised as a transparent and intuitive trend in the bedside monitors. Significance The Sleep State Trend (SST) may provide caregivers a real-time view of sleep state fluctuations and its cyclicity.

Funding Information: This work was supported by the Academy of Finland (Suomen Akatemia): 321235, 313242, 288220, and 310445; Sigrid Jusélius Foundation (Sigrid Juséliuksen Säätiö); Foundation for Pediatric Research (Lastentautien Tutkimussäätiö); and Finnish Brain Foundation (Suomen Aivosäätiö). Publisher Copyright: Copyright © 2022 Tokariev, Oberlander, Videman and Vanhatalo. Up to five percent of human infants are exposed to maternal antidepressant medication by serotonin reuptake inhibitors (SRI) during pregnancy, yet the SRI effects on infants’ early neurodevelopment are not fully understood. Here, we studied how maternal SRI medication affects cortical frequency-specific and cross-frequency interactions estimated, respectively, by phase-phase correlations (PPC) and phase-amplitude coupling (PAC) in electroencephalographic (EEG) recordings. We examined the cortical activity in infants after fetal exposure to SRIs relative to a control group of infants without medical history of any kind. Our findings show that the sleep-related dynamics of PPC networks are selectively affected by in utero SRI exposure, however, those alterations do not correlate to later neurocognitive development as tested by neuropsychological evaluation at two years of age. In turn, phase-amplitude coupling was found to be suppressed in SRI infants across multiple distributed cortical regions and these effects were linked to their neurocognitive outcomes. Our results are compatible with the overall notion that in utero drug exposures may cause subtle, yet measurable changes in the brain structure and function. Our present findings are based on the measures of local and inter-areal neuronal interactions in the cortex which can be readily used across species, as well as between different scales of inspection: from the whole animals to in vitro preparations. Therefore, this work opens a framework to explore the cellular and molecular mechanisms underlying neurodevelopmental SRI effects at all translational levels.

Background Delayed brain function development in small-gestational-age (SGA) infants has been reported. We aimed to quantify rates of immature neonatal EEG patterns and their association with neurodevelopment in SGA full-term neonates. Methods Using a cohort design, 50 SGA (birthweight 3 h. Seventy-three of them were assessed at 2-years-old using Bayley-III-Scales. For EEG analysis, several segments of discontinuous/alternating EEG tracings were selected. Main outcomes measured: (1) Visual analysis (patterns of EEG maturity); (2) Power spectrum in δ, θ, α and β frequency bands; and (3) scores in motor, cognitive and language development. Results (1) SGA infants, compared to AGA, showed: (a) higher percentages of discontinuous EEG, both asynchrony and interhemispheric asymmetry, and bursts with delta-brushes, longer interburst-interval duration and more transients/hour; (b) lower relative power spectrum in δ and higher in α; and (c) lower scores on motor, language and cognitive neurodevelopment. (2) Asymmetry >5%, interburst-interval >5 s, discontinuity >11%, and bursts with delta-brushes >11% were associated with lower scores on Bayley-III. Conclusions In this prospective study, SGA full-term neonates showed high rates of immature EEG patterns. Low-birthweight and immaturity EEG were both correlated with low development scores.

Research using electroencephalography (EEG) as a measure of brain function and maturation has demonstrated links between cortical activity and cognitive processes during infancy and early childhood. The current study examines whether neonatal EEG is correlated with later atypical socioemotional behaviors or neurocognitive delays. Parental-report developmental assessments were administered to families with children ages 24 to 36 months who had previously participated in a neonatal EEG study (N = 129). Significant associations were found between neonatal EEG (higher frequencies in the frontal-polar, temporal, and parietal brain regions) and BITSEA ASD risk scores. Infants with lower EEG power in these brain areas were more likely to have higher risk of socio-emotional problems. When examining sex differences, significant links were found for males but not for females. These results demonstrate some promising associations between early neural biomarkers and later risk for atypical behaviors, which may shape early neurobehavioral development and could lead to earlier identification and intervention.

This study was aimed to assess electrophysiological brain activity in newborns of various gestational ages using electroencephalography (EEG). We summarized the data on the main EEG characteristics of premature infants and their dynamics and described the most common pathological EEG changes that can be identified at this age. We also discussed prognostic value of various neonatal EEG patterns (including their proportion and location), which allow verification of functional brain disorders at early stages. © 2019 ABV-Press Publishing House. All rights reserved. Russian Science Foundation, RSF Финансирование. Работа выполнена при финансовой поддержке гранта №16-18-10371 Российского научного фонда. Исследование прово-дится в рамках программы, поддержанной постановлением №211 Правительства Российской Федерации, соглашение №02.A03.21.0006. Financing. The study was funded by the grant No. 16-18-10 371 from the Russian Science Foundation. The study conducts within a program, supported by the resolution of the Russian Government No. 211, agreement No. 02.A03.21.0006.

openaire: EC/H2020/813483/EU//INFANS Neonatal brain monitoring in the neonatal intensive care units (NICU) requires a continuous review of the spontaneous cortical activity, i.e., the electroencephalograph (EEG) background activity. This needs development of bedside methods for an automated assessment of the EEG background activity. In this paper, we present development of the key components of a neonatal EEG background classifier, starting from the visual background scoring to classifier design, and finally to possible bedside visualization of the classifier results. A dataset with 13,200 5-minute EEG epochs (8–16 channels) from 27 infants with birth asphyxia was used for classifier training after scoring by two independent experts. We tested three classifier designs based on 98 computational features, and their performance was assessed with respect to scoring system, pre- and post-processing of labels and outputs, choice of channels, and visualization in monitor displays. The optimal solution achieved an overall classification accuracy of 97% with a range across subjects of 81–100%. We identified a set of 23 features that make the classifier highly robust to the choice of channels and missing data due to artefact rejection. Our results showed that an automated bedside classifier of EEG background is achievable, and we publish the full classifier algorithm to allow further clinical replication and validation studies.

Neonatal brain monitoring in the neonatal intensive care units (NICU) requires a continuous review of the spontaneous cortical activity, i.e., the electroencephalograph (EEG) background activity. This needs development of bedside methods for an automated assessment of the EEG background activity. In this paper, we present development of the key components of a neonatal EEG background classifier, starting from the visual background scoring to classifier design, and finally to possible bedside visualization of the classifier results. A dataset with 13,200 5-minute EEG epochs (8–16 channels) from 27 infants with birth asphyxia was used for classifier training after scoring by two independent experts. We tested three classifier designs based on 98 computational features, and their performance was assessed with respect to scoring system, pre- and post-processing of labels and outputs, choice of channels, and visualization in monitor displays. The optimal solution achieved an overall classification accuracy of 97% with a range across subjects of 81–100%. We identified a set of 23 features that make the classifier highly robust to the choice of channels and missing data due to artefact rejection. Our results showed that an automated bedside classifier of EEG background is achievable, and we publish the full classifier algorithm to allow further clinical replication and validation studies.

Neural activity is both a driver of brain development and a readout of developmental processes. Changes in neuronal activity are therefore both the cause and consequence of neurodevelopmental compromises. Here, we review the assessment of neuronal activities in both preclinical models and clinical situations. We focus on issues that require urgent translational research, the challenges and bottlenecks preventing translation of biomedical research into new clinical diagnostics or treatments, and possibilities to overcome these barriers. The key questions are (i) what can be measured in clinical settings versus animal experiments, (ii) how do measurements relate to particular stages of development, and (iii) how can we balance practical and ethical realities with methodological compromises in measurements and treatments.

EEG is the gold standard for seizure detection in the newborn infant, but EEG interpretation in the preterm group is particularly challenging; trained experts are scarce and the task of interpreting EEG in real-time is arduous. Preterm infants are reported to have a higher incidence of seizures compared to term infants. Preterm EEG morphology differs from that of term infants, which implies that seizure detection algorithms trained on term EEG may not be appropriate. The task of developing preterm specific algorithms becomes extra-challenging given the limited amount of annotated preterm EEG data available. This paper explores novel deep learning (DL) architectures for the task of neonatal seizure detection in preterm infants. The study tests and compares several approaches to address the problem: training on data from full-term infants; training on data from preterm infants; training on age-specific preterm data and transfer learning. The system performance is assessed on a large database of continuous EEG recordings of 575[Formula: see text]h in duration. It is shown that the accuracy of a validated term-trained EEG seizure detection algorithm, based on a support vector machine classifier, when tested on preterm infants falls well short of the performance achieved for full-term infants. An AUC of 88.3% was obtained when tested on preterm EEG as compared to 96.6% obtained when tested on term EEG. When re-trained on preterm EEG, the performance marginally increases to 89.7%. An alternative DL approach shows a more stable trend when tested on the preterm cohort, starting with an AUC of 93.3% for the term-trained algorithm and reaching 95.0% by transfer learning from the term model using available preterm data. The proposed DL approach avoids time-consuming explicit feature engineering and leverages the existence of the term seizure detection model, resulting in accurate predictions with a minimum amount of annotated preterm data.

Neonates spend most of their life sleeping. During sleep, their brain experiences fast changes in its functional organization. Microstate analysis permits to capture the rapid dynamical changes occurring in the functional organization of the brain by representing the changing spatio-temporal features of the electroencephalogram (EEG) as a sequence of short-lasting scalp topographies—the microstates. In this study, we modeled the ongoing neonatal EEG into sequences of a limited number of microstates and investigated whether the extracted microstate features are altered in REM and NREM sleep (usually known as active and quiet sleep states—AS and QS—in the newborn) and depend on the EEG frequency band. 19-channel EEG recordings from 60 full-term healthy infants were analyzed using a modified version of the k-means clustering algorithm. The results show that ~ 70% of the variance in the datasets can be described using 7 dominant microstate templates. The mean duration and mean occurrence of the dominant microstates were significantly different in the two sleep states. Microstate syntax analysis demonstrated that the microstate sequences characterizing AS and QS had specific non-casual structures that differed in the two sleep states. Microstate analysis of the neonatal EEG in specific frequency bands showed a clear dependence of the explained variance on frequency. Overall, our findings demonstrate that (1) the spatio-temporal dynamics of the neonatal EEG can be described by non-casual sequences of a limited number of microstate templates; (2) the brain dynamics described by these microstate templates depends on frequency; (3) the features of the microstate sequences can well differentiate the physiological conditions characterizing AS and QS. Supplementary Information The online version contains supplementary material available at 10.1007/s10548-021-00861-1.

INTRODUCTION: Perinatal anoxia remains common and can lead to hypoxic-ischemic encephalopathy (HIE), representing a major public health issue. The prognosis of moderate and severe HIE is improved by therapeutic hypothermia (TH) started within 6 hours of life. EEG allows early assessment of HIE severity but its interpretation requires expertise that is not always available.Based on quantitative analysis of neonatal EEGs recorded in the latent phase after anoxia, we aimed to develop a clinical tool for automatic classification of EEG severity and early discrimination of neonates requiring TH. This model will also be tested on EEGs recorded after H6 under TH. Finally, quantitative EEG markers (qEEG) will be used to study the direct fetal brain response to experimental acute anoxia in the fetal sheep.METHODS: 150 EEGs recorded within 6 hours of life after perinatal anoxia were visually graded into 3 severity groups according to the French Classification and quantified using 6 qEEG markers measuring signal amplitude, continuity and frequency content. The qEEG variables that best described the severity groups were combined in machine learning algorithms. Different models were trained on a development cohort (n=90) to classify the EEG severity and then tested on a validation cohort (n=60). The most effective model was also trained to predict clinical outcome at 1 year. This model was then studied on 271 EEGs recorded between H12 and H72 in neonates treated with HT.The same qEEG markers were used in an experimental fetal sheep model to study the normal brain activity in 7 controls and the fetal brain response to acute anoxia induced by repeated umbilical cord occlusions in 9 other fetuses.RESULTS: The qEEG variables significantly discriminated the 3 visual EEG severity groups, as well as between poor or good clinical outcome at 1 year of age. The most effective automatic classification model was based on a SVM algorithm with an accuracy of 80% for EEG severity groups and 94% for clinical outcome.When the model was applied to “TH” EEG recorded after H6, the accuracy was 67% for EEG severity groups.In the experimental part, we developed a new method for in utero EEG acquisition in the fetal sheep. The fetal brain response to acute anoxia was characterized by the suppression of EEG activity during occlusion which would correspond to a neuroprotective adaptive mechanism, and a progressive global alteration of the signal with the severity of the occlusions, which would rather reflect ischemic cerebral repercussion in relation to the failure of cardiovascular adaptation mechanisms. The decrease in SEF during severe occlusions was significantly correlated with neuronal lost.CONCLUSION: The use of clinically relevant qEEG markers allowed to quantitatively describe the severity of post-anoxic neonatal EEGs according to the French classification and to develop an automatic EEG classification model effective in the post-anoxic latent phase to discriminate neonates requiring TH. This model can also be used in neonates treated by TH to automatically assess the evolution of the EEG and thus the long term prognosis. This model can be used to implement a clinical tool for automatic assessment of HIE severity that can assist medical decisions in the first days of life, particularly for TH indication.These qEEG markers reflected the fetal cerebral response to acute anoxia and may represent prenatal markers predictive of cerebral hypoxic-ischemic lesions. In the future, these markers may be used to develop new methods for early detection and prevention of neonatal HIE., INTRODUCTION : L’anoxie périnatale reste fréquente et peut se compliquer d’une encéphalopathie anoxo-ischémique (EAI), représentant un enjeu de santé publique majeur. Le pronostic des EAI modérées et sévères est amélioré par l’hypothermie thérapeutique (HT) débutée avant 6 heures de vie. L’EEG permet d’évaluer précocement la sévérité de l’EAI mais son interprétation nécessite une expertise qui n’est pas toujours disponible.A partir de l’analyse quantitative des EEG néonataux enregistrés en phase de latence post-anoxique, l’objectif de ce travail est de développer un outil de classification automatique de la sévérité des EEG permettant de discriminer précocement les nouveau-nés nécessitant un traitement par HT. Ce modèle sera également testé sur des EEG enregistrés après H6 sous HT. Enfin, les marqueurs quantitatifs EEG (qEEG) seront utilisés pour étudier la réponse cérébrale foetale directe à une anoxie aiguë expérimentale chez le foetus de brebis.METHODES : 150 EEG, enregistrés dans les 6 premières heures de vie après une anoxie périnatale, ont été gradés visuellement en 3 groupes de sévérité selon la classification française et quantifiés à l’aide de 6 marqueurs qEEG mesurant l’amplitude, la continuité et le contenu fréquentiel du signal. Les variables qEEG décrivant au mieux les groupes de sévérité ont été combinées dans des algorithmes d’apprentissage automatique. Différents modèles étaient entraînés à classer la sévérité EEG d’une cohorte de développement (n=90) puis testés sur une cohorte de validation (n=60). Le modèle le plus performant était également entraîné à prédire l’évolution clinique à 1 an puis étudié sur 271 tracés enregistrés après H6 chez des nouveau-nés sous HT.Les mêmes marqueurs qEEG ont été utilisés dans un modèle expérimental de foetus de brebis pour étudier l'activité cérébrale normale in utero chez 7 témoins et la réponse cérébrale foetale à une anoxie aiguë provoquée par des occlusions de cordon ombilical répétées chez 9 autres foetus.RESULTATS : Les variables qEEG discriminaient significativement les 3 groupes de sévérité EEG visuels, ainsi que l’évolution clinique favorable ou défavorable des enfants à l’âge de 1 an. Le modèle de classification automatique le plus efficace était basé sur un algorithme de SVM qui classait correctement 80% des tracés par rapport aux groupes de sévérité visuels et 94% par rapport à l’évolution clinique. 67% des tracés enregistrés après H6 étaient bien classés par ce modèle selon leur groupe de sévérité.Nous avons ensuite développé une nouvelle technique d’acquisition EEG chez le foetus de brebis. La réponse foetale cérébrale à l’anoxie aiguë était caractérisée par une suppression du signal per-occlusion qui semble adaptative, et une altération globale progressive du signal avec la sévérité de l’anoxie qui refléterait la défaillance des mécanismes d’adaptation. La diminution de la SEF lors des occlusions sévères était corrélée à la survenue de lésions cérébrales.CONCLUSION : L’utilisation de marqueurs qEEG pertinents cliniquement et disponibles en routine, a permis de décrire quantitativement la sévérité des tracés et de développer un modèle de classification automatique des EEG efficace dès la phase de latence post-anoxique pour discriminer les nouveau-nés nécessitant un traitement par HT. Ce modèle peut également être utilisé chez les enfants mis en HT pour évaluer automatiquement l’évolution des tracés et donc le pronostic à plus long terme. Il peut servir à l’implémentation d’un outil d’évaluation automatique de la sévérité de l’EAI utilisable au lit du malade comme aide clinique pour les décisions thérapeutiques des premiers jours de vie.L’utilisation de ces marqueurs qEEG a également permis de repérer des marqueurs prénataux potentiellement prédictifs des lésions anoxo-ischémiques cérébrales. Ces marqueurs pourront être utilisés pour le développement de nouveaux moyens de détection précoce et de prévention de l’EAI néonatale.

Electroencephalography (EEG) is an important clinical tool for reviewing sleep-wake cycling in neonates in intensive care. Trace alternant (TA)-a characteristic pattern of EEG activity during quiet sleep in term neonates-is defined by alternating periods of short-duration, high-voltage activity (bursts) separated by lower-voltage activity (inter-bursts). This study presents a novel approach for detecting TA activity by first detecting the inter-bursts and then processing the temporal map of the bursts and inter-bursts. EEG recordings from 72 healthy term neonates were used to develop and evaluate performance of 1) an inter-burst detection method which is then used for 2) detection of TA activity. First, multiple amplitude and spectral features were combined using a support vector machine (SVM) to classify bursts from inter-bursts within TA activity, resulting in a median area under the operating characteristic curve (AUC) of 0.95 (95% confidence interval, CI: 0.93 to 0.98). Second, post-processing of the continuous SVM output, the confidence score, was used to produce a TA envelope. This envelope was used to detect TA activity within the continuous EEG with a median AUC of 0.84 (95% CI: 0.80 to 0.88). These results validate how an inter-burst detection approach combined with post processing can be used to classify TA activity. Detecting the presence or absence of TA will help quantify disruption of the clinically important sleep-wake cycle., 4 pages, to be appearing in upcoming 2020 EMBC Conference

Many preterm infants suffer from neural disorders caused by early birth complications. The detection of children with neurological risk is an important challenge. The electroencephalogram is an important technique for establishing long-term neurological prognosis. Within this scope, the goal of this study is to propose an automatic detection of abnormal preterm babies’ electroencephalograms (EEG). A corpus of 316 neonatal EEG recordings of 100 infants born after less than 35 weeks of gestation were preprocessed and a time series of standard deviation was computed. This time series was thresholded to detect Inter Burst Intervals (IBI). Temporal features were extracted from bursts and IBI. Feature selection was carried out with classification in one step so as to select the best combination of features in terms of classification performance. Two classifiers were tested: Multiple Linear Regressions and Support Vector Machines (SVM). Performance was computed using cross validations. Methods were validated on a corpus of 100 infants with no serious brain damage. The Multiple Linear Regression method shows the best results with a sensitivity of 86.11% ± 10.01%, a specificity of 77.44% ± 7.62% and an AUC (Area under the ROC curves) of 0.82 ± 0.04. An accurate detection of abnormal EEG for preterm infants is feasible. This study is a first step towards an automatic analysis of the premature brain, making it possible to lighten the physician’s workload in the future.

Background Electroencephalography (EEG) monitoring is widely used in neonatal intensive care units (NICUs). However, conventional EEG report generation processes are time-consuming and labor-intensive. Therefore, an automatic, objective, and comprehensive pipeline for brain age estimation and EEG report conclusion prediction is urgently needed to assist clinician's decision-making. Methods We recruited patients who underwent EEG monitoring from the NICU at Children's Hospital of Fudan University from Jan. 2016 to Mar. 2018. A total of 1,851 subjects were enrolled, including the patient's conceptional age (CA) and the clinical EEG report conclusion (normal, slightly abnormal, moderately abnormal, or severely abnormal). A total of 1,591 subjects were used to generate predictive models and 260 were used as the validation dataset. We developed Auto-Neo-EEG (an automatic prediction system to assist clinical neonatal EEG report generation), including signal feature extraction, supervised machine learning realized by gradient boosted models, to estimate brain age and predict EEG report conclusion. Results The predicted results from the validation dataset were compared with the clinical observations to assess the performance. In the independent validation dataset, the model could achieve accordance 0.904 on estimating brain age for neonates with normal clinical EEG report conclusion, and differences between the predicted and observed brain age were strongly related with EEG report conclusion abnormality. Further, as for the EEG report conclusion prediction, the model could achieve area under the curve (AUC) of 0.984 for severely abnormal situations, and 0.857 for moderately abnormal ones. Conclusions The Auto-Neo-EEG has the high accuracy of estimating brain age and EEG report conclusion, which can potentially greatly accelerate the EEG report generation processes assist in clinical decision making.

In this paper, we propose a new method, based on the nonlinear energy operator (NLEO), to automatically detect burst---suppression (B---S) patterns in multichannel newborn electroencephalograms (EEGs). The proposed approach consists of two algorithms: (1) per-brain region B---S detection and (2) global B---S detection. At first, B---S patterns are detected in each channel using NLEO. Average of NLEO values obtained for all the channels is then calculated to detect the presence of B---S patterns in each brain region. After local B---S detection, the global B---S detection algorithm classifies a sample-point as burst if most of regions are bursting. Otherwise, the sample-point is classified as suppression. The proposed method is validated using a database composed of multichannel EEG signals acquired from 6 neonates. The experimental results show that the proposed approach can detect bursts which occur locally and classify global B---S patterns with a very high accuracy of 98%.

Electroencephalography (EEG) is used in neonatal care to assess encephalopathy, seizure recognition and classification, to make epilepsy syndrome diagnoses and to assess the maturity of neonatal brain activity. A basic understanding of the EEG is very helpful in ensuring that clinicians gain as much information as possible from this helpful, non-invasive investigation. The neonatal EEG is complex and accurate reporting requires detailed clinical information to be provided on request forms. Even when this is provided EEG reports are frequently returned to the neonatal unit loaded with technical details, making it difficult for neonatal staff to fully understand them. This article reviews the basics of EEG, the changes seen with increasing gestational age, and changes seen in common pathologies. We also provide a structured approach to the interpretation of the neonatal EEG report, and discuss its role in prognostication. Amplitude integrated EEG is reviewed in our companion paper.

The effect of the complete electrode model on electroencephalography forward and inverse computations is explored. A realistic neonatal head model, including a skull structure with fontanels and sutures, is used. The electrode and skull modeling differences are analyzed and compared with each other. The results suggest that the complete electrode model can be considered as an integral part of the outer head model. To achieve optimal source localization results, accurate electrode modeling might be necessary., This paper investigates finite element method-based modeling in the context of neonatal electroencephalography (EEG). In particular, the focus lies on electrode boundary conditions. We compare the complete electrode model (CEM) with the point electrode model (PEM), which is the current standard in EEG. In the CEM, the voltage experienced by an electrode is modeled more realistically as the integral average of the potential distribution over its contact surface, whereas the PEM relies on a point value. Consequently, the CEM takes into account the subelectrode shunting currents, which are absent in the PEM. In this study, we aim to find out how the electrode voltage predicted by these two models differ, if standard size electrodes are attached to a head of a neonate. Additionally, we study voltages and voltage variation on electrode surfaces with two source locations: 1) next to the C6 electrode and 2) directly under the Fz electrode and the frontal fontanel. A realistic model of a neonatal head, including a skull with fontanels and sutures, is used. Based on the results, the forward simulation differences between CEM and PEM are in general small, but significant outliers can occur in the vicinity of the electrodes. The CEM can be considered as an integral part of the outer head model. The outcome of this study helps understanding volume conduction of neonatal EEG, since it enlightens the role of advanced skull and electrode modeling in forward and inverse computations. NEW & NOTEWORTHY The effect of the complete electrode model on electroencephalography forward and inverse computations is explored. A realistic neonatal head model, including a skull structure with fontanels and sutures, is used. The electrode and skull modeling differences are analyzed and compared with each other. The results suggest that the complete electrode model can be considered as an integral part of the outer head model. To achieve optimal source localization results, accurate electrode modeling might be necessary.