This thesis deals with analysis and processing of the Sleep Electroencephalogram (EEG) signals. The scope of this thesis can be split into several areas. The first area is application of the Independent Component Analysis (ICA) method for EEG signal analysis. A model of EEG signal formation is proposed and conditions under which this model is valid are examined. It is shown that ICA can be used to remove non-deterministic artifacts contained in the EEG signals. The second area of interest is analysis of stationarity of the Sleep EEG signal. Methods to identify stationary signal segments and to analyze statistical properties of these stationary segments are presented. The third area of interest focuses on spectral analysis of the Sleep EEG signals. Analyses are performed that shows the processes that form particular parts of EEG signals spectrum. Also, random signals that are an integral part of the EEG signals analysis are performed. The last area of interest focuses on elimination of the transition processes that are caused by the filtering of the short EEG signal segments.

Healthy ageing has been found to be accompanied by changes in slow wave activity (SWA) and cognitive function. Furthermore, these changes have been seen predominantly in the prefrontal cortex (PFC) compared to other regions of the cortex. Current theories of cognitive ageing propose that this occurs due to a specified deterioration of neuronal substrates of the PFC, and as such, changes in SWA and cognitive function may decline at similar rates due to similar underlying aetiology. The main aim of the current thesis was to explore age-related differences in electroencephalographic (EEG) SWA during the first NREM period and cognitive performance that relies on the integrity of the PFC: executive function and social cognition. The extent to which executive function (reliant on dorsolateral PFC areas) and social cognitive function (reliant on ventromedial PFC regions) show similar age-related deterioration was investigated in Study 1. Here, 16 young (22.2 years) and 16 older (71.5 years) adults were administered with a cognitive testing battery including executive function measures: Verbal Fluency (VF) and Tower of London (TOL); as well as measures of social cognition: Go/No-go, Emotional Prosody and Ekman 60 Faces. Not all measures of PFC function were affected to the same extent. The older group performed significantly worse on the TOL, but not on the VF test. Additionally, simple aspects of social cognition did not display differences between the groups, but the older group performed significantly worse than the young group on more complex aspects of recognition of emotion from facial expression (Ekman 60 Faces) and Emotional Prosody. As most studies of cognitive ageing are cross-sectional and show large agerelated changes, the remainder of this thesis focused on age-related changes using a longitudinal design over a relatively small ageing period (mean = 6.29 years). The average age of participants at baseline was 67.1 years and the average age at follow-up was 73.4 years. In Study 2, in a sample of 11 participants, performance on executive function tests was measured (TOL, VF and Wisconsin Card Sorting Test: WCST). As found in the cross-sectional analyses reported in Study 1, the TOL task was found to be the most sensitive indicator of age-related changes, as this showed a decline with age; whereas, VF and WCST remained stable over time. Furthermore, in Study 3, localised SWA was recorded via EEG, and significant declines were found in low frequency delta (0.5 – 1 Hz), which was localised to the left frontal region.

Le somnambulisme est une parasomnie commune, caractérisée par des éveils incomplets lors des stades de sommeil lent, au cours desquels les individus atteints présentent des comportements moteurs d’une complexité variable accompagnés de confusion et d’un jugement altéré. La littérature actuelle suggère que ce trouble serait associé à des particularités de l’activité en ondes lentes et des oscillations lentes, deux indices de l’intégrité du processus homéostatique et de la profondeur du sommeil. Toutefois, en raison de certaines lacunes méthodologiques dans les études existantes, le rôle de ces marqueurs électroencéphalographiques dans la pathophysiologie du somnambulisme reste à éclaircir. Notre premier article a donc investigué d’éventuelles anomalies de l’activité en ondes lentes et des oscillations lentes chez les somnambules, en comparant leur sommeil au cours de la nuit entière à celui de participants contrôles. De plus, comme les somnambules semblent réagir différemment (en termes de fragmentation du sommeil notamment) des dormeurs normaux à une pression homéostatique accrue, nous avons comparé l’activité en ondes lentes et les oscillations lentes en nuit de base et suite à une privation de sommeil de 38 heures. Les résultats de nos enregistrements électroencéphalographiques chez 10 somnambules adultes et neuf participants contrôles montrent une élévation de la puissance spectrale de l’activité en ondes lentes et de la densité des oscillations lentes en nuit de récupération par rapport à la nuit de base pour nos deux groupes. Toutefois, contrairement à plusieurs études précédentes, nous ne n’observons pas de différence entre somnambules et dormeurs normaux quant à l’activité en ondes lentes et aux oscillations lentes pour aucune des deux nuits. Au-delà ce certaines considérations méthodologiques ayant pu contribuer à ce résultat inattendu, nous croyons qu’il justifie un questionnement sur l’hétérogénéité des somnambules comme population. Notre deuxième article s’est penché sur les facteurs électroencéphalographiques transitoires susceptibles d’être associés au déclenchement des épisodes de somnambulisme. Nous avons comparé les fluctuations d’activité en ondes lentes et des oscillations lentes dans les minutes avant des épisodes de somnambulisme spontanés (c.a.d.: non associés à un stimulus identifiable) à celles survenant avant des éveils normaux comparables chez 12 somnambules adultes. Nous montrons que, comparativement aux éveils normaux, les épisodes de somnambulisme sont précédés d’un sommeil plus profond, tel qu’indiqué par une plus grande densité spectrale de l’activité en ondes lentes et une plus grande densité des oscillations lentes. Cet approfondissement du sommeil, spécifique aux épisodes de somnambulisme, semble survenir sur un laps de temps relativement long (>3 minutes), et non abruptement au cours des secondes précédant l’épisode. Ces données ouvrent un questionnement quant aux mécanismes en jeu dans la survenue des épisodes de somnambulisme spontanés. Globalement, cette thèse suggère que des phénomènes liés à l’activité en ondes lentes et aux oscillations lentes seraient liés au déclenchement des épisodes de somnambulisme, mais que des études supplémentaires devront être menées afin de délimiter le rôle précis que ces marqueurs jouent dans la pathophysiologie du somnambulisme. / Sleepwalking is a common parasomnia characterized by sudden but incomplete arousals out of non-rapid eye movement sleep during which predisposed individuals display motor behaviours of various complexity, accompanied by mental confusion and altered judgement. A growing body of evidence suggests that this condition could be associated with atypical patterns in slow wave activity and slow oscillations, both markers of the integrity of the homeostasis process and of sleep intensity. However, due to methodological limitations in past studies, the role of these electroencephalographic markers in the pathophysiology of sleepwalking remains unclear. Our first article aimed at describing slow wave activity and slow oscillations abnormalities in sleepwalkers by comparing whole night sleep in 10 adult sleepwalkers and 9 control participants. In addition, since past studies have shown that increased homeostatic pressure has differential effects on sleepwalkers versus normal controls (e.g., in terms of sleep fragmentation), we compared slow wave activity and slow oscillations during baseline sleep and recovery sleep after 38 hours of sleep deprivation in patients and controls. Results show that sleep deprivation increases slow wave activity power density and slow oscillations density in both groups. However, contrary to our predictions, no group differences were noted on any of the two nights on slow wave activity or slow oscillations. Beyond methodological considerations which may partially account for this unexpected result, this study opens questions as to the homogeneity of sleepwalkers as a clinical population. Our second study focused on transient electroencephalographic fluctuations that may be associated with the onset of sleepwalking episodes. We compared slow wave activity and slow oscillations fluctuations in the moments leading up to spontaneous (that is, occurring without an identifiable internal or external stimuli) somnambulistic episodes recorded in the sleep laboratory in 12 adult sleepwalkers and comparing these patterns to those observed prior to non-behavioural awakenings observed in the same patients. We showed that when compared to non-behavioural awakenings from the same sleep stage and sleep period, somnambulistic episodes were preceded by deeper sleep, as indicated by higher slow wave activity power density and slow oscillations density. This deepening of sleepwalkers’ sleep occurs over a relatively long period of time (>3 minutes) before the episode, rather than abruptly in the seconds preceding episode onset. These findings raise key questions about fundamental mechanisms involved in the occurrence of spontaneously recorded somnambulistic episodes. Taken as a whole, the results from the work presented in this thesis show that electrophysiological processes related to slow wave activity and slow oscillations play a role in the occurrence of somnambulistic episodes. However, the functional significance of these electroencephalographic markers in the pathophysiology of sleepwalking remains to be clarified.

Die schlafendokrine Regulation wird massgeblich beeinflusst durch Hirnareale, die auch für die Pathophysiologie psychiatrischer, und insbesondere affektiver Störungen, eine wichtige Rolle spielen. Dazu zählen neben dem Hypothalamus, und insbesondere dem paraventrikulären Kerngebiet (PVN), auch limbische Areale wie der Hippokampus und die Amygdala, der präfrontale Kortex sowie die afferenten Kerngebiete im Hirnstamm, insbesondere der Locus coeruleus (LC) und die Raphe Kerne. Der Schlaf und die nächtliche Hormonsekretion können als dynamische Prozesse begriffen werden, die gekennzeichnet sind durch eine fein abgestimmte Abfolge neuronaler Aktivitäten in bestimmten Kerngebieten im Verlauf einer Nacht. Somit ermöglichen polysomnographische und endokrine Untersuchungen Aussagen über mögliche Funktionsänderungen im Zusammenspiel dieser verschiedenen Kerngebiete und leisten daher einen wichtigen Beitrag für die (neuro-) psychiatrische Forschung. In der folgenden Arbeit haben wir insbesondere die dynamischen Aspekte der Schlafarchitektur und die nächtliche Hormonsekretion als Parameter verwendet, um verschiedene Einflussfaktoren auf die schlafendokrine Regulation zu untersuchen. Dabei war es mein Ziel, neben der Diagnose Major Depression auch eine Interaktion zwischen Geschlecht, Alter und Symptomatologie der Depression zu untersuchen. Wir haben dargestellt, dass das Geschlecht einen massgeblichen Einfluss auf diejenigen Schlafparameter ausübt, die bisher als charakteristische Merkmale einer Major Depression angesehen wurden. Weiterhin konnten wir zeigen, dass das Alter (bzw. die Menopause bei Frauen) die schlafendokrinen Veränderungen im Rahmen einer Major Depression geschlechtsspezifisch moduliert. Auch beobachteten wir, dass die klinischen Merkmale einer Major Depression durch geschlechtsspezifische Faktoren beeinflusst werden, wie auch neuere Studien nahelegen. Insbesondere stellen wir die Hypothese auf, dass sogenannte atypische Symptome der Depression bei jüngeren Frauen gehäuft vorkommen und in einer Unteraktivität der HPA Achse (vielleicht als Ausdruck einer genetischen Untererregbarkeit), der afferenten, aktivierenden Serotoninneurone und der Amygdala Kerne begründet liegen. An dieser Konstellation könnte auch eine chronische, aber moderate inflammatorische Immunreaktion beteiligt sein. Im Gegensatz dazu weisen Patienten mit nicht atypischer Depression Zeichen einer Überaktivität der HPA Achse und der Amygdala Kerne auf, die sich im Sinne eines Teufelskreises gegenseitig verstärken und afferente katecholaminerge Neurone einbeziehen können. Die Möglichkeit, dass eine durch inflammatorische Zytokine hervorgerufene Sensibilisierung der HPA Aktivität zu dieser Entwicklung beiträgt scheint plausibel und sollte in zukünftigen Studien überprüft werden. Die unmittelbare Relevanz dieser Überlegungen und Befunde wird unterstrichen durch geschlechtsspezifische Behandlungserfolge bei Patienten mit Major Depression und durch unsere Befunde einer geschlechtsspezifischen Beeinflussung der schlafendokrinen Regulation durch Peptidhormone. Schliesslich haben wir in einer Gruppe von Patientinnen mit Multipler Sklerose (MS) gezeigt, dass unabhängig von den Symptomen einer Major Depression, die hochdosierte Gabe von Glukokortikoiden depressions-typische Veränderungen der schlafendokrinen Regulation induzieren kann. Da diese Patientinnen neben der akuten Glukokortikoidgabe auch eine akute inflammatorische Reaktion (akuter MS Schub) aufwiesen, stellt sich die Frage nach einer Interaktion zwischen der inflammatorischen Reaktion und der Hypothalamus-Hypophysen-Nebennierenrinden (HPA) Achse. Einerseits können immunologische Prozesse den Schlaf beeinflussen, und andererseits gibt es zunehmend Hinweise für einen Zusammenhang zwischen affektiven Symptomen und Immunreaktionen. Da das Geschlecht Immunreaktionen wesentlich beeinflusst, könnten die geschlechtsspezifischen Unterschiede hinsichtlich der klinischen Symptome und der schlafendokrinen Veränderungen im Rahmen einer Major Depression Ausdruck unterschiedlicher zugrundeliegender immunologischer Mechanismen sein. Unterschiede in der Pathophysiologie depressiver Störungen eröffnen die Möglichkeit, die Therapie depressiver Patienten spezifischer auf die zugrundeliegenden Pathomechanismen auszurichten. Dadurch könnten erstens die Behandlung optimiert und zweitens neue spezifischere Behandlungsstrategien für Patienten mit affektiven Störungen entwickelt werden. / The sleep-endocrine regulation is critically influenced by brain areas, which also play an important role for the pathophysiology of psychiatric, and particularly affective disorders. These include the hypothalamus, and particularly the paraventricular nucleus (PVN), but also limbic areas such as the hippocampus and the amygdala, the prefrontal cortex as well as afferent brainstem nuclei, in particular the locus coeruleus (LC) and the raphe nuclei. Sleep and nocturnal hormone secretion can be viewed as dynamic processes, which are characterised by a fine-tuned pattern of neuronal activity in certain brain nuclei in the course of a night. Thus, polysomnographic and endocrine investigations give us insight into possible functional changes in the interplay of these different nuclei and hence provide an important contribution to the field of (neuro-) psychiatric research. In the following piece of work we have used in particular parameters which highlight the dynamic aspects of the sleep-architecture and hormone secretion to examine the various modulators of sleep-endocrine regulation. My aim was to investigate, besides the effect of the diagnosis of major depression, also the interaction of gender, ageing and clinical features of major depression. We have demonstrated that gender has a critical influence on sleep-endocrine changes, which so far have been considered typical of major depression. Also, we could show that ageing (respectively the menopause in women) affects sleep-endocrine changes in patients with major depression in a gender-specific manner. Furthermore, we noted that the clinical features of major depression are also influenced by gender-specific factors, as suggested by recent studies. In particular, we put forward the hypothesis that so-called atypical features of depression are common in young female patients and reflect hypoactivity of the HPA axis (possibly due to a genetic hyposensitivity to stimulation), the afferent activating serotonin neurones and the amygdala nuclei. A chronic but moderate inflammatory immune reaction could also play a role in this situation. Conversely, patients with non-atypical depression show signs of overactivity of the HPA axis and the amygdala nuclei, which can be mutually reinforced, leading to a vicious circle and which can include afferent catecholaminergic neurones. The possibility that priming of the HPA axis with inflammatory cytokines contributes to this development seems plausible and should be examined in future studies. The immediate relevance of these considerations and data is highlighted by gender-specific treatment responses among patients with major depression and our own data demonstrating gender-specific modulation of sleep-endocrine regulation by peptide hormones. Finally, we have shown in a group of female patients with multiple sclerosis (MS) that high doses of glucocorticoids can induce depression-like changes of sleep-endocrine regulation, independently of clinical symptoms of depression. As these patients were treated with high doses of glucocorticoids and suffered from an acute inflammatory reaction (acute MS relapse), the question arises about an interaction between the inflammatory reaction and the hypothalamo-pituitary-adrenocortical (HPA) axis. On the one hand immunological processes can affect sleep and on the other hand there is a growing number of studies suggesting an association between affective symptoms and immune reactions. As gender critically influences immune reactions, the gender-specific differences in clinical features and sleep-endocrine changes in patients with major depression could reflect underlying differences in immunological mechanisms. Differences regarding the pathophysiology of major depression open up the possibility to adjust treatments more specifically to the underlying pathophysiology. Thereby, one could firstly optimise the therapy and secondly develop new and more specific treatment strategies for patients with affective disorders.

Οι υπνικές άτρακτοι είναι απότομες αλλαγές της ρυθμικής δραστηριότητας που χαρακτηρίζονται από σταδιακή αύξηση και κατόπιν μείωση του πλάτους. Εμφανίζονται κυρίως στα στάδια 2,3 και 4 του υπνικού εγκεφαλογράμματος. Τοπογραφικές αναλύσεις έχουν δείξει την ύπαρξη δύο ξεχωριστών τύπων υπνικών ατράκτων, «αργές» και «ταχείς», περίπου στα 12 και 14 Hz, αντίστοιχα. Υπάρχουν ενδείξεις ότι υπάρχουν τουλάχιστον δύο, λειτουργικά, ξεχωριστές γεννήτριες υπνικών ατράκτων, που αντιστοιχούν στις κλάσεις συχνοτήτων. Ο λόγος της εργασίας αυτής ήταν η επεξεργασία υπνικών ατράκτων με την τεχνική Ανάλυσης Ανεξάρτητων Συνιστωσών (ICA) με σκοπό την έρευνα της πιθανότητας εξαγωγής, στα από την ICA ανακατασκευαζόμενα ηλεκτροεγκεφαλογραφήματα (ΗΕΓ), «συνιστωσών» ατράκτων που αντιστοιχούν σε ξεχωριστές δομές ΗΕΓ, και η μελέτη των πηγών που δημιουργούν αυτές τις συνιστώσες. Χρησιμοποιήθηκαν 8κάναλες καταγραφές υπνικών ατράκτων ΗΕΓ από έναν εξεταζόμενο, που καταγράφηκαν στα πλαίσια του Biopattern Network of Excellence, οι οποίες αρχικά επεξεργάστηκαν με ένα φίλτρο FIR με συχνότητες αποκοπής (-3dB) στα 6 και 21 Hz. Κατόπιν εφαρμόστηκε η ανάλυση ICA και εξάχθηκαν οι ανεξάρτητες συνιστώσες (ICs). Έγινε επιλογή των συνιστωσών οι οποίες θα ανακατασκεύαζαν τα ΗΕΓ και τέλος ανακατασκευάσθηκαν αυτά. Στα ανακατασκευασμένα ΗΕΓ έφαρμόστηκε η ανάλυση LORETA. Πρωτού γίνει όμως αυτό έγινε μία εξομείωση του «ευθύ» και «ανάστροφου» προβλήματος. Αυτό έγινε για να μελετήσουμε κατά πόσον θα μπορούσαμε να εξάγουμε αξιόπιστα αποτελέσματα από την τεχνική LORETA με τόσο μικρό αριθμό καναλιών καταγραφής. Η μελέτη αυτή έδειξε ότι τα αποτελέσματά μας θα μπορούσαν να μας δώσουν αξιόπιστες πληροφορίες όσον αφορά την ευρεία περιοχή παραγωγής των ατράκτων και όχι την ακριβή τους θέση. Τα τελικά αποτελέσματα έδειξαν ότι υπάρχει διαφοροποίηση, όσον αφορά την περιοχή παραγωγής τους, και σταθερότητα των πηγών που σχετίζονται με συνιστώσες ατράκτων που ανακατασκευάζονται από ξεχωριστές ομάδες ανεξαρτήτων συνιστωσών (ICs). / Sleep spindles are bursts of rhythmic activity characterized by progressively increasing, then gradually decreasing amplitude, present predominantly in stages 2, 3 and 4 of the sleep electroencephalogram (EEG). Topographic analyses of sleep spindle incidence suggested the existence of two distinct sleep spindle types, “slow” and “fast” spindles at approximately 12 and 14 Hz respectively. There are indications that there exist at least two functionally separated spindle generators, corresponding to each frequency spectrum class. The purpose of the present study was to process sleep spindles with Independent Component Analysis (ICA) in order to investigate the possibility of extracting, in the ICA-reconstructed EEG, spindle “components” corresponding to separate EEG activity patterns, and to investigate the sources underlying these spindle components. We used 8-channel EEG recordings of sleep spindles of a single subject, recorded in the framework of the Biopattern Network of Excellence, which were processed by a FIR filter with cut-off frequencies (-3 dB) at 6 and 21 Hz. Afterwards, ICA was applied and ICs were extracted. There were a choice of the ICs which would reconstruct the EEG and the EEG were finally reconstructed. Source analysis using Low-Resolution Brain Electromagnetic Tomography (LORETA) was applied on the reconstructed EEGs. Before that we made a simulation of the “direct” and “inverse” problem. This was made in order to investigate if we would extract reliable results from the LORETA technique with only 8-channel recordings. The investigation stated that the results could give reliable information only for the brain sites at which the spindle generators were located and not for their exact position. Results indicate separability and stability of sources related to sleep spindle components reconstructed from separate groups of Independent Components (ICs).

Diverses études se sont penchées sur les paramètres EEG du sommeil en ondes lentes, y compris l’activité en ondes lentes en lien avec le somnambulisme, mais les résultats se révèlent inconsistants et contradictoires. Le premier objectif de la présente étude était d’analyser quantitativement l’EEG en sommeil en mesurant les fluctuations de puissance spectrale en delta (1-4 Hz) et delta lent (0.5-1 Hz) avant des épisodes de somnambulisme. Le second était de détecter les oscillations lentes (> 75 μV, fréquence d'environ 0.7-0.8 Hz) et très lentes (> 140 μV, fréquence d'environ 0.7-0.8 Hz) afin d'examiner leur changement d'amplitude et de densité avant de tels épisodes. Suite à une privation de sommeil de 25 heures, les enregistrements polysomnographiques de 22 adultes atteints de somnambulisme ont été scrutés. L’analyse des 200 secondes avant les épisodes révèle que ceux-ci ne sont pas précédés d’une augmentation graduelle de puissance spectrale en delta ni en delta lent, tant sur les dérivations frontale, centrale que pariétale. Toutefois, une hausse statistiquement significative de la densité des oscillations lentes et des oscillations très lentes a été observée au cours des 20 sec immédiatement avant le début des épisodes. Reste à déterminer le rôle exact de ces paramètres de l’EEG en sommeil par rapport à la manifestation et au diagnostic des parasomnies en sommeil lent. / Several studies have investigated slow-wave sleep EEG parameters, including slow-wave activity (SWA) in relation to somnambulism, but results have been both inconsistent and contradictory. The first goal of the present study was to conduct a quantitative analysis of sleepwalkers’ sleep EEG by studying fluctuations in spectral power for delta (1-4 Hz) and slow delta (0.5-1 Hz) before the onset of somnambulistic episodes. A secondary aim was to detect slow wave oscillations to examine their changes in amplitude and density prior to behavioral episodes of somnambulism. Twenty-two adult sleepwalkers were investigated polysomnographically following 25 h of sleep deprivation. Analysis of patients’ sleep EEG over the 200 sec prior to the episodes’ onset revealed that the episodes were not preceded by a gradual increase in spectral power for either delta or slow delta over frontal, central, or parietal leads. However, time course comparisons revealed significant changes in the density of slow and very slow wave oscillations, with significant increases occurring during the final 20 sec immediately preceding episode onset. The specificity of these sleep EEG parameters for the occurrence and diagnosis of NREM parasomnias remains to be determined.

Made available in DSpace on 2014-12-17T14:13:20Z (GMT). No. of bitstreams: 1 CarlosNA.pdf: 2422732 bytes, checksum: f5d47defbd8e5b492fe5d2975c8286ee (MD5) Previous issue date: 2006-12-28 / Innumerable studies have focused been reported on the sleep spindles (SS), Sharp Vertex Waves (SVW) and REM, NREM Sleep as indicators interpreting EEG patterns in children. However, Frequency and Amplitud Gradient (FAG) is rarely cited sleep parameter in children,that occurs during NREM Sleep. It was first described by Slater and Torres, in 1979, but has not been routinely evaluated in EEG reports. The aim of this study was to assess the absence of SS, SVW and FAG, as an indication of neurological compromise in children. The sample consisted of 1014 EEGs of children referred to the Clinical Neurophysiology Laboratory, Hospital Universit?rio de Bras?lia (HUB), from January 1997 to March 2003, with ages ranging from 3 months to 12 years old, obtained in spontaneous sleep or induced by choral hydrate. The study was transversal and analytical, in which, visual analysis of EEG traces was perfumed individually and independently by two electroencephalographers without prior knowledge of the EEG study or neurological findings. After EEG selection, the investigators analyzed the medical reports in order to define and correlate neurological pattern was classified according to the presence or absence of neurological compromise, as Normal Neurological Pattern (NNP), and Altered Neurological Pattern (ANP) respectively. From the visual analysis of the EEG(s), it was possible to characterize 6 parameters: 1- FAG present (64,1%); 2- FAG absent (35,9%); 3 - normal SS (87,9%); 4 - altered SS s (12,1%); 5 - normal SVW s (95,7%); 6 - altered SVW s (4,3%). The prevalence of well-formed FAG is found in the 3 months to 5 years age group in the children with NNF. FAG was totally absent from the age of 10 years. When comparing the three sleep graphielements, it was observed that SVW and SS were predominant in children with NNF. However, FAG absent was more prevalent in the ANF than in altered SS an SVW. The statistical analysis showed that there is a strong association of FAG absent, with isolated alteration, in ANF patients, in that the prevalence ratio was 6,60. The association becomes stronger when FAG absent + altered SS(s) is considered (RP= 6,68). Chi-square test, corrected by Yates technique, showed a highly significant relation for FAG &#961;= 0,00000001, for error X of 5%, or else the 95% confidence interval (&#961;<0,05). Thus, the FAG absent were more expressive in ANF patient than altered SS(s) and SVW(s). The association becomes stronger in order to establish a prognostic relation, when the FAG is combined with the SS. The results os this study allow us to affirm that the FAG, when absent at ages ranging from 3 months to 5 years , is an indication of neurological compromise. FAG is an age-dependent EEG parameter and incorporated systematically, in the interpretation criteria of the EEG of children s sleep, not only in the maturational point of view, but also neurological disturbances with encephalic compromise / In?meras pesquisas t?m focalizado periodicamente os fusos do sono (FS), as ondas agudas do v?rtex (OAV), o complexo K e o padr?o do sono REM e NREM como indicadores de avalia??o eletrencefalogr?fica da inf?ncia. O GFA ? um padr?o EEG do sono de crian?as que ocorre durante o sono NREM, raramente citado na literatura, e que, descrito pela primeira vez por Slater e Torres, em 1979, e n?o devidamente valorizado na rotina dos laudos EEG. Nas montagens referenciais ? caracterizado por uma progressiva diminui??o de voltagem e aumento de freq??ncia na dire??o p?stero-anterior. O objetivo desta tese, foi analisar o gradiente de freq??ncia e amplitude, um padr?o EEG do sono de crian?as que ocorre durante o sono NREM; estudar os fusos do sono(FS), ondas agudas do v?rtex (OAV), como indicadores de comprometimento neurol?gico. A popula??o de estudo constitui-se de 1014 EEG de crian?as atendidas no Laborat?rio de Neurofisiologia Cl?nica do Hospital Universit?rio de Bras?lia (HUB) no per?odo de janeiro de 1997 a mar?o de 2003, nas faixas et?rias de 3 meses a 12 anos de idade, obtidos em sono espont?neo ou induzido por hidrato de cloral. O tipo de ensaio foi transversal-anal?tico, onde os EEG foram avaliados independentemente por 2 examinadores sem pr?vio conhecimento do padr?o neurol?gico e da indica??o cl?nica. Ap?s an?lise visual do EEG, foram pesquisados os prontu?rios m?dicos de todas as crian?as inclu?das no estudo, a fim de definir e associar o padr?o neurol?gico com os par?metros fusos do sono, ondas agudas do v?rtex, e GFA. O padr?o neurol?gico foi classificado segundo a presen?a ou aus?ncia de comprometimento neurol?gico em padr?o neurol?gico normal (PNN) e padr?o neurol?gico anormal (PNA), respectivamente. Com base na an?lise visual dos EEG, foi caracterizado o GFA em duas categorias: 1) GFA presente (64,1%); 2) GFA ausente ( 35,9%); 3) FS normais (87,9%); 4) FS alterados (12,1%); 5) OAV normais (95,7%); 6) OAV alteradas (4,3%). A melhor express?o do GFA presente com PNN ocorreu nas faixas et?rias de 3 meses a 5 anos. Observou-se tamb?m que o GFA torna-se ausente a partir dos 10 anos de idade em crian?as com PNN. Comparando os 3 grafoelementos do sono, as OAV e FS, foram respectivamente predominantes nas crian?as com PNN. A an?lise estat?stica mostrou que existe uma forte associa??o de aus?ncia de GFA, como altera??o isolada, nos pacientes com PNA, uma vez que a raz?o de preval?ncia foi de 6,60. A associa??o torna-se mais forte, quando se considerou GFA ausente + FS alterados (RP=11,9) e GFA ausente + FS alterados + OAV alterados (RP= 6,68). O teste do qui-quadrado, com corre??o pela t?cnica de Yates, mostrou uma rela??o altamente significativa, quando envolvido o GFA. Assim, o GFA ausente foi mais expressivo no PNA que os FS e as OAV alterados. A associa??o se torna ainda mais forte a ponto de estabelecer uma rela??o com valor progn?stico, quando o GFA se encontra-se combinado com o FS. Embora estatisticamente significante, n?o houve associa??o quando as OAV se encontra isoladas em compara??o aos demais par?metros. Os dados obtidos neste estudo permitem afirmar que, entre os grafo- elementos do sono, o GFA, quando ausente nas faixas et?rias de 3 meses a 5 anos, ? um indicador de comprometimento neurol?gico, sendo tal conclus?o mais expressiva do que os par?metros FS e OAV. O GFA ? um par?metro idade dependente e deve ser valorizado e incorporado, sistematicamente, aos crit?rios de interpreta??o do tra?ado EEG do sono de crian?as, tanto no ponto de vista da avalia??o maturacional como dos dist?rbios neurol?gicos com comprometimento encef?lico