Search

The presence of a respiratory-related cortical activity during tidal breathing is abnormal and a hallmark of respiratory difficulties, but its detection requires superior discrimination and temporal resolution. The aim of this study was to validate a computational method using EEG covariance (or connectivity) matrices to detect a change in brain activity related to breathing. In 17 healthy subjects, EEG was recorded during resting unloaded breathing (RB), voluntary sniffs, and breathing against an inspiratory threshold load (ITL). EEG were analyzed by the specially developed covariance-based classifier, event-related potentials, and time-frequency (T-F) distributions. Nine subjects repeated the protocol. The classifier could accurately detect ITL and sniffs compared with the reference period of RB. For ITL, EEG-based detection was superior to airflow-based detection ( P < 0.05). A coincident improvement in EEG-airflow correlation in ITL compared with RB ( P < 0.05) confirmed that EEG detection relates to breathing. Premotor potential incidence was significantly higher before inspiration in sniffs and ITL compared with RB ( P < 0.05), but T-F distributions revealed a significant difference between sniffs and RB only ( P < 0.05). Intraclass correlation values ranged from poor (−0.2) to excellent (1.0). Thus, as for conventional event-related potential analysis, the covariance-based classifier can accurately predict a change in brain state related to a change in respiratory state, and given its capacity for near “real-time” detection, it is suitable to monitor the respiratory state in respiratory and critically ill patients in the development of a brain-ventilator interface.

Older age produces numerous changes in cognitive processes, including slowing in the rate of mental processing speed. There has been controversy over the past three decades about whether this slowing is generalized or process-specific. A growing literature indicates that it is process-specific and suggests it is most dramatic at the interface where a stimulus input is translated into a response output. We tested this hypothesis using a task in which young and older adult males made either compatible or incompatible responses to the word LEFT or RIGHT shown briefly and variously located in a 4 row × 6 column matrix surrounded by # signs or by letters chosen randomly from the sets A-G or A-Z. Processing speed was measured using P300 latency and reaction time. Experimental effects on these two measures provided support for the hypothesis in revealing that stimulus identification processes were preserved, whereas processes related to translating a stimulus input into a designated response output and then selecting that response were compromised in the elderly.

Epilepsy, a neurological disorder in which patients suffer from recurring seizures, affects approximately 1% of the world population. In spite of available drug and surgical treatment options, more than 25% of individuals with epilepsy have seizures that are uncontrollable. For these patients with intractable epilepsy, the unpredictability of seizure occurrence underlies an enhanced risk of sudden unexpected death or morbidity. A system that could warn the patient of the impending event or trigger an antiepileptic device would dramatically increase the quality of life for those patients. Here, we proposed a patient-specific algorithm for possible seizure warning using machine learning classification of 34 algorithmic features derived from EEG–ECG recordings. We evaluated our algorithm on unselected and continuous recordings of 12 patients (total of 108 seizures and 3178-h). Good out-of-sample performances were observed around 25% of the patients with an average preictal period around 30 min and independently of the EEG type (scalp or intracranial). Inspection of the most discriminative EEG–ECG features revealed that good classification rates reflected specific physiological precursors, particularly related to certain stages of sleep. From these observations, we conclude that our algorithmic strategy enables a quantitative way to identify “pro-ictal” states with a high risk of seizure generation.

We studied brain activity during the displacement of attention in a modified visuo-spatial orienting paradigm. Using a behaviorally relevant no-shift condition as a control, we asked whether ipsi- or contralateral parietal alpha band activity is specifically related to covert shifts of attention. Cue-related event-related potentials revealed an attention directing anterior negativity (ADAN) contralateral to the shift of attention and P3 and contingent negative variation waveforms that were enhanced in both shift conditions as compared to the no-shift task. When attention was shifted away from fixation, alpha band activity over parietal regions ipsilateral to the attended hemifield was enhanced relative to the control condition, albeit with different dynamics in the upper and lower alpha subbands. Contralateral-to-attended parietal alpha band activity was indistinguishable from the no-shift task.

Abtract Line Garnero passed away in the night between July 12 and 13, 2009 after an intense but too short a battle against cancer. Line was one of the most active scientific figures of the medical imaging community. A major motivation to Line has always been to facilitate communication and mutual understanding amongst her collaborators, to help them build strong multidisciplinary partnerships and projects. In that respect, she was a visionary of the evolution in the organization and conceptions of academic research for multidisciplinary research. She has always been a strong proponent of an open and translational approach to research in applied science, within her group, her laboratory and as a Deputy Director of the new Research Center of the Brain and Spinal Chord Institute which has been inaugurated just recently on the campus of La Salpetriere Hospital in Paris (France). This research center illustrates the multidisciplinary approach that Line cherished and nurtured all along her career: 400 scientists are now sharing the same environment and workspace to work on all aspects of neuroscience research: from genes to behavior and neuroimaging. This short text is a tribute to Line, and an attempt from some of her closest collaborators, to summarize her major contributions. Download : Download full-size image

The modern analysis of multivariate electrical brain signals requires advanced statistical tools to automatically extract and quantify their information content. These tools include machine learning techniques and information theory. They are currently used both in basic neuroscience and challenging applications such as brain computer interfaces. We review here how these methods have been used at the Laboratoire d’Electroencephalographie et de Neurophysiologie Appliquee (LENA) to develop a general tool for the real time analysis of functional brain signals. We then give some perspectives on how these tools can help understanding the biological mechanisms of information processing.

Resume Une methodologie basee sur une analyse en composantes independantes spatiale (ACI) a ete developpee specifiquement pour des donnees d’electroencephalographie (EEG) pour extraire des reseaux fonctionnels a large echelle. Les resultats preliminaires sont presentes dans cet article. Les donnees EEG ont ete enregistrees chez 15 sujets effectuant une tâche visuelle. Elles ont ete pretraitees et reconstruites sur les cortex de chaque sujet, grâce a une methode de norme minimale. Une ACI spatiale a ete realisee sur la puissance de ces signaux reconstruits et a permis d’en extraire des composantes independantes. Une interpolation sur une surface commune a ete effectuee pour proceder a une analyse de groupe. Une technique de regroupement des composantes a fait ressortir les composantes les plus reproductibles entre les sujets qui, regroupees, definissent des reseaux fonctionnels d’interet. Pour evaluer ces resultats, les reseaux extraits en IRMf sur les memes sujets ont ete projetes sur le cortex grâce a une methode de projection d’image volumique sur une surface utilisant un diagramme de Voronoi. Une comparaison visuelle montre que des reseaux EEG et des reseaux IRMf sont proches. Les perspectives de ce travail preliminaire prevoient de definir une regression entre ces reseaux pour analyser plus pertinemment les reseaux fonctionnels en EEG et acquerir l’expertise pour les etiqueter.

Recently, the neuroscience field took a particular interest in the use of a neuroimaging technique called ‘hyperscanning’. This new technique consists in the simultaneous recording of the hemodynamic or neuroelectric activities of multiple subjects. Behind this small technical step lays a giant methodological leap. Groundbreaking insight in the understanding of social cognition shall be achieved if the right paradigms are implemented. A growing number of studies demonstrate the potential of this recent technique. In this paper, we will focus on current issues and future perspectives of brain studies using hyperscanning. We will also add to this review two studies initiated by Line Garnero. These studies will illustrate the promising possibilities offered by hyperscanning through two different key phenomena pertaining to social interaction: gesture imitation and joint attention.

International audience; OBJECTIVE: Tracking the level of performance in cognitive tasks may be useful in environments, such as aircraft, in which the awareness of the pilots is critical for security. In this paper, the usefulness of EEG for the prediction of performance is investigated. METHODS: We present a new methodology that combines various ongoing EEG measurements to predict performance level during a cognitive task. We propose a voting approach that combines the outputs of elementary support vector machine (SVM) classifiers derived from various sets of EEG parameters in different frequency bands. The spectral power and phase synchrony of the oscillatory activities are used to classify the periods of rapid reaction time (RT) versus the slow RT responses of each subject. RESULTS: The voting algorithm significantly outperforms classical SVM and gives a good average classification accuracy across 12 subjects (71%) and an average information transfer rate (ITR) of 0.49bit/min. The main discriminating activities are laterally distributed theta power and anterio-posterior alpha synchronies, possibly reflecting the role of a visual-attentional network in performance. CONCLUSIONS: Power and synchrony measurements enable the discrimination between periods of high average reaction time versus periods of low average reaction time in a same subject. Moreover, the proposed approach is easy to interpret as it combines various types of measurements for classification, emphasizing the most informative. SIGNIFICANCE: Ongoing EEG recordings can predict the level of performance during a cognitive task. This can lead to real-time EEG monitoring devices for the anticipation of human mistakes.

Interrupting a focal, chronic infusion of GABA to the rat motor cortex initiates the progressive emergence of a sustained spiking electroencephalographic (EEG) activity, associated with myoclonic jerks of the corresponding body territory. This activity is maintained over several hours, has an average frequency of 1.5 Hz, is localized to the infusion site and never generalizes. The GABA withdrawal syndrome (GWS) has therefore features of partial status epilepticus. Changes in EEG signals associated with the GWS were studied in freely moving rats by measuring the phase synchrony between bilateral epidural records from the neocortex. Our results showed (i) epileptic activity was associated with a striking decrease in phase synchrony between all pairs of electrodes including the focus, predominantly in the 1–6 Hz frequency range. There was a mean decrease of 75.34±5.26% in phase synchrony levels between the period before GABA interruption and the period after epileptic activity appeared. (ii) This reduction in synchrony contrasted with an increase of power spectral density in the corresponding EEG channels over the same 1–6 Hz frequency range, (iii) neither changes in synchrony nor in nonlinear dynamics were detected before the first EEG spikes, (iv) systemic injection of ketamine, an antagonist of N-methyl- d -aspartic acid (NMDA) receptors, modified transiently both epileptic activity and the synchrony profile. (v) Spiking activity and synchrony changes were suppressed by reperfusion of GABA. Our data suggest that, during a partial status epilepticus, interactions between the epileptic focus and connected neocortical neuronal populations are dramatically decreased in low frequencies.

In an effort to find quantitative markers of attention during reading, we performed simultaneous MEG and intracranial EEG recordings of alpha, beta and gamma band oscillatory activity in an epileptic patient during attentive vs. inattentive reading. We compared the strength and spatial organization of those oscillations between the two conditions. Time-frequency analysis of intracranial EEG revealed an attention-dependant network of gamma band energy increases and alpha/beta energy suppressions in response to word presentation. MEG data were also characterized by power increases in the gamma band and power suppressions in the alpha/beta range, mirroring the intracranial effects. The agreement we found between MEG and intracranial EEG power modulations suggests that MEG can detect attention-related modulations of cortical oscillatory processes and thus provide a reliable non-invasive measure of attention during reading in individual subjects.

The present study investigates the predictability of a subject's state based on the classification of the underlying brain activity recorded via magnetoencephalography (MEG). We use Second Order Blind Identification (SOBI) to reduce the high dimensionality of MEG sensors into a smaller number of task-related components. A classification of distinct cognitive states is then achieved by feeding the spectral power of these components into a Support Vector Machine (SVM). We tested this approach on data from one subject during a visuomotor control experiment and found that our method outperforms classification based on the spectral powers computed directly from the MEG sensor array. Our findings suggest that combining SOBI and SVM may provide a reliable classifier for the prediction of cognitive states in MEG.

In experimental synchronization studies a continuous phase variable is commonly estimated from a scalar time series by means of its representation on the complex plane. The aim is to obtain a pair of functions [A(t), phi(t)] defining its instantaneous amplitude and phase, respectively. However, any arbitrary pair of functions cannot be considered as the amplitude and the phase of the real observable. Here, we point out some criteria that the pair [A(t), phi(t)] must observe to unambiguously define the instantaneous amplitude and phase of the observed signal. In this work, we illustrate how the complex representation may fail if the signal possesses a multi-component or a broadband spectra. We also point out a practical procedure to test whether a signal, not displaying a single oscillation at a unique frequency, has a narrow-band behavior. Implications for the study of phase interdependencies are illustrated and discussed. Phase dynamics estimated from electric brain activities recorded from an epileptic patient are also discussed.

Imaging the dynamics of distributed phase synchrony across brain signals is of crucial importance for the study of large-scale interactions in the brain, and requires combining at the same time, wide coverage of the brain with high spatial and temporal resolution. Electro- and magneto-encephalography (EEG-MEG), which provide full head coverage measurements of the human brain activity, can potentially satisfy those needs. Nevertheless, EEG-MEG signals reflects the integration of separately generated brain processes on the scalp that typically overlap in space and give rise to spurious phase-locking between their signals. Moreover, current phase synchronization measures do not have a sufficient time resolution to capture very brief periods of phase locking between brain signals, because of their dependence on a window of time integration. We present here a new, non-invasive technique for characterizing the phase synchronization between brain regions at high spatial and temporal resolution. An efficient inverse problem algorithm was used to estimate, from the MEG signals and with the help of the anatomical MRI, the corresponding intracranial brain sources on the cortical surface. The synchronization analysis was then directly performed on the cortex by the characterization of common instantaneous frequencies between groups of cortical sources which preserve a fine temporal resolution. The proposed method was illustrated by its application to MEG data recorded during absence seizures in two epileptic patients. The technique visualizes local and short-lasting synchronization patterns leading to the seizure, thus providing new potential for understanding non-invasively the origin of epileptic discharges.

Objective There is accumulated evidence that mesial temporal lobe seizures are preceded by a preictal transition that evolves over minutes to hours. In the present study, we investigated these possible preictal changes in long-term intracranial recordings of five patients by a measure of phase synchronization. In order to clearly distinguish preictal changes from all the other interictal states, we developed an automatic extraction of representative patterns of interictal synchronization activity. This reference library was used to classify the successive synchronization patterns of long-term recordings into groups of similar patterns. Altered states of brain synchronization were identified as deviating from patterns in the reference library and were evaluated relative to the times of seizure onset in terms of sensitivity and specificity. Methods A phase-locking measure was estimated using a sliding window analysis on 15 frequency bands (2Hz steps between 0 and 30Hz), for all pairs of EEG channels in the epileptogenic temporal lobe (14–20 channels), over the entire data sets (total analyzed duration 305h). The preictal identification encompasses three basic stages: (1) a preprocessing stage involving the determination of a reference library of characteristic interictal synchronization patterns using a K -means algorithm, and the identification of discriminant variables differentiating interictal from preictal states, (2) a classification stage of the synchronization pattern via a minimum Mahalanobis distance to the reference patterns, as well as detection of outliers, (3) an evaluation stage of the sensitivity and specificity of the detection by receiver-operating characteristic curves. Results In most of the cases (36 of 52 seizures, i.e. 70%), a specific state of brain synchronization can be observed several hours before the actual seizure. The changes involved both increases and decreases of the synchronization levels, occurring mostly within the 4–15Hz frequency band, and were often localized near the primary epileptogenic zone. Conclusions The analysis of phase synchronization offers a way to distinguish between a preictal state and normal interictal activity. These findings suggest that brain synchronizations are preictally altered in the epileptogenic temporal lobe, inducing a pathological state of higher susceptibility for seizure activity. Significance Phase synchronization is capable of extracting information from the EEG that allow the definition of a preictal state. Although the proposed analysis does not constitute genuine seizure anticipation, these changes in neuronal synchronization may provide helpful information for prospective seizure warning.

The mechanisms underlying the transition of brain activity toward epileptic seizures remain unclear. Based on nonlinear analysis of both intracranial and scalp electroencephalographic (EEG) recordings, different research groups have recently reported dynamical smooth changes in epileptic brain activity several minutes before seizure onset. Such preictal states have been detected in populations of patients with mesial temporal lobe epilepsy (MTLE) and, more recently, with different neocortical partial epilepsies (NPEs). In this paper, we are particularly interested in the spatio-temporal organization of epileptogenic networks prior to seizures in neocortical epilepsies. For this, we characterize the network of two patients with NPE by means of two nonlinear measures of interdependencies. Since the synchronization of neuronal activity is an essential feature of the generation and propagation of epileptic activity, we have analyzed changes in phase synchrony between EEG time series. In order to compare the phase and amplitude dynamics, we have also studied the degree of association between pairs of signals by means of a nonlinear correlation coefficient. Recent findings have suggested changes prior to seizures in a wideband frequency range. Instead, for the examples of this study, we report a significant decrease of synchrony in the focal area several minutes before seizures (/spl Gt/30 min in both patients) in the frequency band of 10-25 Hz mainly. Furthermore, the spatio-temporal organization of this preictal activity seems to be specifically related to this frequency band. Measures of both amplitude and phase coupling yielded similar results in narrow-band analysis. These results may open new perspectives on the mechanisms of seizure emergence as well as the organization of neocortical epileptogenic networks. The possibility of forecasting the onset of seizures has important implications for a better understanding, diagnosis and a potential treatment of the epilepsy.

This paper introduces the use of wavelet analysis to follow the temporal variations in the coupling between oscillatory neural signals. Coherence, based on Fourier analysis, has been commonly used as a first approximation to track such coupling under the assumption that neural signals are stationary. Yet, stationary neural processing may be the exception rather than the rule. In this context, the recent application to physical systems of a wavelet-based coherence, which does not depend on the stationarity of the signals, is highly relevant. This paper fully develops the method of wavelet coherence and its statistical properties so that it can be practically applied to continuous neural signals. In realistic simulations, we show that, in contrast to Fourier coherence, wavelet coherence can detect short, significant episodes of coherence between non-stationary neural signals. This method can be directly applied for an ‘online’ quantification of the instantaneous coherence between two signals.

The study of dynamic changes in neural activity preceding epileptic seizure allows the characterization of a preictal state several minutes before seizure onset. This opens up new perspectives for studying the mechanisms of epileptogenesis as well as for possible therapeutic interventions, which represent a major breakthrough. In this review the authors present and discuss the results from their group in this domain using nonlinear analysis of brain signals, as well as the limitations of this topic and current questions.

Summary Background New methods derived from non-linear analysis of intracranial recordings permit the anticipation of an epileptic seizure several minutes before the seizure. Nevertheless, anticipation of seizures based on standard scalpelectroencephalographical (EEG) signals has not been reported yet. The accessibility to preictal changes from standard EEGs is essential for expanding the clinical applicability of these methods. Methods We analysed 26 scalp-EEG/video recordings, from 60 min before a seizure, in 23 patients with temporal-lobe epilepsy. For five patients, simultaneous scalp and intracranial EEG recordings were assessed. Long-term changes before seizure onset were identified by a measure of non-linear similarity, which is very robust in spite of large artifacts and runs in real-time. Findings In 25 of 26 recordings, measurement of non-linear changes in EEG signals allowed the anticipation of a seizure several minutes before it occurred (mean 7 min). These preictal changes in the scalp EEG correspond well with concurrent changes in depth recordings. Interpretation Scalp-EEG recordings retain sufficient dynamical information which can be used for the analysis of preictal changes leading to seizures. Seizure anticipation strategies in real-time can now be envisaged for diverse clinical applications, such as devices for patient warning, for efficacy of ictal-single photon emission computed tomography procedures, and eventual treatment interventions for preventing seizures.

This paper introduces a new method, single-trial phase locking statistics (S-PLS) to estimate phase locking in single trials of brain signals between two electrodes. The possibility of studying single trials removes an important limitation in the study of long-range synchrony in brain signals. S-PLS is closely related to our previous method, phase locking statistics (PLS) that estimates phase locking over a set of trials. The S-PLS method is described in detail and applied to human surface recordings during the task of face-recognition. We compare these results with those provided by PLS and show that they are qualitatively very similar, although S-PLS provides better discrimination of synchronic episodes.

This paper studies gamma-band responses from two implanted epileptic patients during a simple visual discrimination task. Our main aim was to ascertain, in a reliable manner, whether evoked (stimulus-locked) and induced (triggered by, but not locked to, stimuli) responses are present in intracranial recordings. For this purpose, we introduce new methods adapted to detect the presence of gamma responses at this level of recording, intermediary between EEG-scalp and unicellular responses. The analysis relies on a trial-by-trial time-frequency analysis and on the use of surrogate data for statistical testing. We report that visual stimulation reliably elicits evoked and induced responses in human intracranial recordings. Induced intracranial gamma activity is significantly present in short oscillatory bursts (a few cycles) following visual stimulation. These responses are highly variable from trial to trial, beginning after 200 ms and lasting up to 500 ms. In contrast, intracranial-evoked gamma responses concentrate around 100 ms latencies corresponding to evoked responses observed on the scalp. We discuss our results in relation to scalp gamma response in a similar protocol [Tallon-Baudry et al. (1996) J. Neurosci., 16, 4240-4249] and draw some conclusions for bridging the gap between gamma oscillations observed on the scalp surface and their possible cortical sources.

This article presents, for the first time, a practical method for the direct quantification of frequency‐specific synchronization (i.e., transient phase‐locking) between two neuroelectric signals. The motivation for its development is to be able to examine the role of neural synchronies as a putative mechanism for long‐range neural integration during cognitive tasks. The method, called phase‐locking statistics (PLS), measures the significance of the phase covariance between two signals with a reasonable time‐resolution (

Epileptic seizures are a principal brain dysfunction with important public health implications, as they affect 0.8% of humans. Many of these patients (20%) are resistant to treatment with drugs1. The ability to anticipate the onset of seizures in such cases would permit clinical interventions. The view of chronic focal epilepsy now is that abnormally discharging neurons act as pacemakers to recruit and entrain other normal neurons by loss of inhibition and synchronization into a critical mass2. Thus, pre-ictal changes should be detectable during the stages of recruitment. Traditional signal analyses, such as the count of focal spike density3, the frequency coherence4 or spectral analyses are not reliable predictors. Non-linear indicators may undergo consistent changes around seizure onset5,6,7. Our objective was to follow the transition into seizure by reconstructing intracranial recordings in implanted patients as trajectories in a phase space and then introduce non-linear indicators to characterize them8,9. These indicators take into account the extended spatio–temporal nature of the epileptic recruitment processes10 and the corresponding physiological events governed by short-term causalities in the time series. We demonstrate that in most cases (17 of 19), seizure onset could be anticipated well in advance (between 2–6 minutes beforehand), and that all subjects seemed to share a similar 'route' towards seizure.

We studied subdural recordings from a patient with an unusually focal and stable occipito-temporal epileptic discharge under four experimental conditions. The series of time intervals between successive spike discharges displayed a few (3-5) clusters of periodic values representing statistically significant short-term periodicities when tested against surrogate data. This short-term predictability was modulated during the different experimental conditions by periodicity shifts of the order of 15-30 ms. Correspondingly, there was an increased gamma-band (30-70 Hz) coherence between the epileptic focus and surrounding recording sites. We conclude that the focal epileptic activity is part of an extended network of neural activities which exert a fast modulation reflected in changes of transiently periodic activities.

We report here on a first attempt to settle the methodological controversy between advocates of two alternative reconstruction approaches for temporal dynamics in brain signals: the single-channel method (using data from one recording site and reconstructing by time-lags), and the multiple-channel method (using data from a spatially distributed set of recordings sites and reconstructing by means of spatial position). For the purpose of a proper comparison of these two techniques, we computed a series of EEG-like measures on the basis of well-known dynamical systems placed inside a spherical model of the head. For each of the simulations, the correlation dimension estimates obtained by both methods were calculated and compared, when possible, with the known (or estimated) dimension of the underlying dynamical system. We show that the single-channel method fails to reliably quantify spatially extended dynamics, while the multichannel method performs better. It follows that the latter is preferable, given the known spatially distributed nature of brain processes. Hum. Brain Mapping 5:26-47, 1997. (c) 1997 Wiley-Liss, Inc.

A fundamentally important problem for cognitive psychophysiologists is selection of the appropriate off-line digital filter to extract signal from noise in the event-related brain potential (ERP) recorded at the scalp. Investigators in the field typically use a type of finite impulse response (FIR) filter known as moving average or boxcar filter to achieve this end. However, this type of filter can produce significant amplitude diminution and distortion of the shape of the ERP waveform. Thus, there is a need to identify more appropriate filters. In this paper, we compare the performance of another type of FIR filter that, unlike the boxcar filter, is designed with an optimizing algorithm that reduces signal distortion and maximizes signal extraction (referred to here as an optimal FIR filter). We applied several different filters of both types to ERP data containing the P300 component. This comparison revealed that boxcar filters reduced the contribution of high-frequency noise to the ERP but in so doing produced a substantial attenuation of P300 amplitude and, in some cases, substantial distortions of the shape of the waveform, resulting in significant errors in latency estimation. In contrast, the optimal FIR filters preserved P300 amplitude, morphology, and latency and also eliminated high-frequency noise more effectively than did the boxcar filters. The implications of these results for data acquisition and analysis are discussed.

Although considerable information on cellular and network mechanisms of epilepsy exists, it is still not understood why, how, and when the transition from interictal to ictal state takes place. The authors review their work on nonlinear EEG analysis and provide consistent evidences that dynamical changes in the neural activity allows the characterization of a preictal state several minutes before seizure onset. This new neurodynamical approach of ictogenesis opens new perspectives for studying the basic mechanisms in epilepsy as well as for possible therapeutic interventions.

The transition of brain activity towards an epileptic seizure is still a poorly understood phenomenon. Dynamic changes in brain activity have been detected several minutes before seizure emergence in populations of patients with mesial temporal lobe epilepsy (MTLE), using non-linear analysis of intracranial EEG recordings. Similar detection of a pre-ictal state has been obtained with standard scalp EEG recordings using a modified non-linear method. Here we applied this strategy to the seizures of patients with neocortical partial epilepsy. Results obtained by non-linear similarity analysis of 41 seizures from 11 patients with refractory partial epilepsy originating from various sites of the neocortex showed that (i) a pre-ictal state was detected in 90% of the patients and in 83% of the seizures whatever their location, with a mean anticipation time of 7.5 min; (ii) similar pre-ictal dynamic changes were detected when non-linear analysis methods were applied to either intracranial or scalp EEG recordings; (iii) the recording sites exhibiting these pre-ictal changes were distributed both within the epileptogenic focus and at remote locations; (iv) most pre-ictal dynamic changes were not correlated with linear changes in the frequency spectrum or with changes in the visually inspected EEG and the patients' behaviour. Hypotheses on the neuronal mechanisms underlying the pre-ictal period are discussed. The present results, together with those recently obtained in an MTLE population, suggest that changes in pre-ictal dynamics are a general phenomenon associated with seizure emergence in a wide population of patients with partial epilepsy, wherever the epileptogenic focus is located. The possibility of anticipating the onset of seizures has considerable therapeutic implications.

We address the problem of detecting, from scalar observations, the time scales involved in synchronization of complex oscillators with several spectral components. Using a recent data-driven procedure for analyzing nonlinear and nonstationary signals [Huang, Proc. R. Soc. London A 454, 903 (1998)], we decompose a time series in distinct oscillation modes which may display a time varying spectrum. When applied to coupled oscillators with multiple time scales, we found that motions are captured in a finite number of phase-locked oscillations. Further, in the synchronized state distinct phenomena as phase slips, anti-phase or perfect phase locking can be simultaneously observed at specific time scales. This fully data-driven approach (without a priori choice of filters or basis functions) is tested on numerical examples and illustrated on electric intracranial signals recorded from an epileptic patient. Implications for the study of the build-up of synchronized states in nonstationary and noisy systems are pointed out.