Congedo et al. (2008). Thus, it is important to remove noise in order to en- hance signal, and to perform feature extraction in order to feed the classifica- tion algorithm with relevant features.
Several methods based on Independent Component Analysis have been proposed to enhance the Signal to Noise Ratio (SNR) and to remove arte- facts. However, these methods are not specifically designed to separate brain activities and they are supervised. Indeed, after decomposition in different components, it is necessary to select (manually or thanks to spatio-temporal prior) components containing evoked potentials. In this work, Event-RelatedPotentials (ERPs) are considered and an unsupervised denoising method is used. It is based on the xDAWN algorithm Rivet et al. (2009), which has been specifically conceived to maximize the SNR of ERPs.
Auditory evoked potentials and their clinical significance
Auditory evoked potentials (AEPs) are a subclass of event- relatedpotentials (ERPs). ERPs are defined as brain responses which are time-locked to some event, such as a sensory stimulus. Averaged ERPs are thought to originate from synchronous activity in pyramidal cells in the activated areas. ERPs result mainly from the summation of cortical excitatory and inhibitory post-synaptic potentials triggered by the release of neurotransmitters such as GABA and glutamate into the synaptic cleft . Recent neurophysiological evidence supports the notion that the features of ERPs result from activity in several cortical sources that are intrinsically connected . The change in amplitude of the AEPs in response to various sound pressure levels (SPLs) is referred to as loudness dependence of auditory evoked potentials (LDAEP) , and is considered as a measure of serotonergic activity. [25–28] Moreover, there are hints of the influence of other neurotrans- mitters such as dopamine and nitric oxide [29,30]. Literature suggests that a pronounced LDAEP of the N1/P2 components reflects low central serotonergic neurotransmission . The inverse relationship between LDAEP and central serotonergic activity has been shown by different methods and in different psychiatric disorders [27,31–35].
Keywords: biomedical signal processing, spatial filtering, asymptotical performance analysis, brain-computer interface, P300
In cognitive neuroscience, it is useful to explore brain activ- ity through evoked potentials (EP) or event-relatedpotentials (ERP) recorded by electro-encephalography (EEG), e.g. [1, 2]. For instance, ERPs allow to investigate i) the basic functional pathways through early ERPs or EPs as auditory, visual or so- matosensory networks, and ii) cognitive pathways through late ERPs which are more related to memory tasks, execution of attention and emotion. ERP experiments usually involve the presentation of several kinds of stimuli and suppose that there exists a typical spatio-temporal pattern which is time-locked to each kind of stimuli (also called events).
The P3 Component as a Measure of Resource Allocation?
A method for contrasting these theoretical positions is EEG, a noninvasive technique of recording brain activity from electrodes placed on the participant’s scalp. Event- relatedpotentials (ERPs) are generated by averaging over segments of EEG activity time-locked to an exter- nally generated event. The averaging process increases the observable signal by removing ongoing non-time- locked EEG activity, which is treated as background noise. The resulting ERP waveform contains a number of positive and negative deflections, which are referred to as ERP components. The P3 component of the ERP occurs 300–600 msec poststimulus presentation and is evoked most strongly by a rare event among a sequence of frequent items, through so called oddball tasks. In the context of rapid serial visual presentation (RSVP), the P3 has been argued to be a correlate of working memory update ( Vogel et al., 1998).
Improving our ability to detect conscious processing in non communicating patients remains a major goal of clinical cognitive neurosciences. In this perspective, several functional brain imaging tools are currently under development. Bedside cognitive event-relatedpotentials (ERPs) derived from the EEG signal are a good candidate to explore consciousness in these patients because: 1) they have an optimal time resolution within the millisecond range able to monitor the stream of consciousness, 2) they are fully non-invasive and relatively cheap, 3) they can be recorded continuously on dedicated individual systems to monitor consciousness and to communicate with patients, 4) and they can be used to enrich patients’ autonomy through brain-computer interfaces. We recently designed an original auditory rule extraction ERP test that evaluates cerebral responses to violations of temporal regularities that are either local in time, or global across several seconds. Local violations led to an early response in auditory cortex, independent of attention or the presence of a concurrent visual task, while global violations led to a late and spatially distributed response that was only present when subjects were attentive and aware of the violations. In the present work, we report the results of this test in 65 successive recordings obtained at bedside from 49 non-communicating patients affected with various acute or chronic neurological disorders. At the individual level, we confirm the high specificity of the ‘global effect’: only conscious patients presented this proposed neural signature of conscious processing. Here, we also describe in details the respective neural responses elicited by violations of local and global auditory regularities, and we report two additional ERP effects related to stimuli expectancy and to task learning, and we discuss their relations to consciousness.
Objective: Abnormal event-relatedpotentials (ERP) and slowing of the waking electroencephalographic (EEG) activity have been reported in patients with obstructive sleep apneas (OSA). This study aimed at evaluating whether an association exists between the severity of ERP abnormalities and EEG slowing in order to better understand cerebral dysfunction in OSA.
Event-relatedpotentials (ERP) have been proposed to improve the differential diagnosis of non-responsive patients. We investigated the potential of the P300 as a reliable marker of conscious processing in patients with locked-in syndrome (LIS). Eleven chronic LIS patients and 10 healthy subjects (HS) listened to a complex-tone auditory oddball paradigm, first in a passive condition (listen to the sounds) and then in an active condition (counting the deviant tones). Seven out of nine HS displayed a P300 waveform in the passive condition and all in the active condition. HS showed statistically significant changes in peak and area amplitude between conditions. Three out of seven LIS patients showed the P3 waveform in the passive condition and five of seven in the active condition. No changes in peak amplitude and only a significant difference at one electrode in area amplitude were observed in this group between conditions. We conclude that, in spite of keeping full consciousness and intact or nearly intact cortical functions, compared to HS, LIS patients present less reliable results when testing with ERP, specifically in the passive condition. We thus strongly recommend applying ERP paradigms in an active condition when evaluating consciousness in non-responsive patients.
Neurological signals are generally very weak in amplitude and strongly noisy. As a result, one of the major challenges in neuroscience is to be able to eliminate noise and thus exploit the maximum amount of information contained in neurological signals (EEG, ERP, Evoked Potentials, ...). In our project, we aim to highlight the ERP's N400 wave which the behavior, the amplitude and the latency may reflect the effects of vowelling and semantic priming in Arabic language. For that reason, we consider a nonlinear filtering method based on discrete 10th order Daubechies discrete wavelet transform combined to principal component analysis, to improve the quality of the recorded ERP signals. Thus, among all tested wavelets, the Daubechies one allows a significant improvement of the signal to noise ratio while using only 10 ERP trials. In addition, we compare and illustrate the effectiveness of this method to that obtained using the averaging technique implemented on EEGLab toolbox. In a second step, the Mexican Hat function have been used to achieve continuous wavelet analysis of the filtered signals. This method permits us to get an alternative representation of the ERPs and to detect avec more accuracy the N400 wave.
FIGURE 2 | Global field power (GFP) of survivors according to their global effect (GE)(+/–) and local effect (LE)(+/–) status. GFP was computed from mean event-relatedpotentials (ERPs) of deviant (red) and standard (blue) conditions and was plotted with a confidence interval at 95% (shaded areas), respectively, in GE+ patients (N = 40; see upper left panel), GE- patients (N = 103; see lower left panel), LE+ patients (N = 86; see upper right panel), and LE- patients (N = 57; see lower right panel). This figure is only shown to illustrate mean ERP patterns: (i) responses to each of the five sounds, (ii) contingent negative variation (CNV) component visible as a ramping ongoing slope in particular in the GE+ group, (iii) mismatch negativity (MMN)-P3a for LE+, and (iv) late P3b component in the GE+ group. No statistical test was calculated given that patients were selected for the corresponding category at the single-subject level statistics.
Keywords: EEG, ERP, ERS, ERSP, single-trial, baseline, additive model, multiplicative gain model
Electroencephalography and magnetoencephalography methods have become standard tools to study brain mechanisms. Different approaches have been used to unveil brain electrical activity in rela- tion to sensory, motor, or cognitive events using electrical potential variations recorded either at the scalp level or from intra-cranial electrodes. The study of changes of the ongoing electroencephalo- gram (EEG) in response to stimulation started with event-relatedpotentials (ERP) techniques, which relies on measuring the ampli- tude and latency of post-stimulus peaks in stimulus-locked EEG trial averages. The standard ERP model relies on the hypothe- sis that ERPs consist of stereotyped patterns of stimulus-locked electrical activity, superimposed onto an independent stationary stochastic EEG processes ( Basar and Dumermuth, 1982 ; Luck, 2005 ; Nunez and Srinivasan, 2006 ). In the ERP model, every single- trial contains a noisy version of the grand average ERP, and, when averaging trials, “stationary” or “non-time-locked” background EEG elements of the signal cancel out.
The aim of this work was to investigate the mechanisms that shape evoked electroencephalographic (EEG) and magneto-encephalographic (MEG) responses. We used a neuronally plausible model to characterise the dependency of response components on the models parameters. This generative model was a neural mass model of hierarchically arranged areas using three kinds of inter-area connections (forward, backward and lateral). We investigated how responses, at each level of a cortical hierarchy, depended on the strength of connections or coupling. Our strategy was to systematically add connections and examine the responses of each successive architecture. We did this in the context of deterministic responses and then with stochastic spontaneous activity. Our aim was to show, in a simple way, how event-related dynamics depend on extrinsic connectivity. To emphasise the importance of nonlinear interactions we tried to disambiguate the components of event-relatedpotentials (ERPs) or event-related fields (ERFs) that can be explained by a linear superposition of trial-specific responses and those engendered nonlinearly (e.g. by phase-resetting).
Scalp voltages with a cap of 64 electrodes distributed according to the 10-20 system and simultaneous electro-oculograms were recorded (linked-earlobe referenced). For each condition, 100 epochs were recorded with an ISI of 2,100 msec (400 msec pre-stimulus). Data were corrected for eye movements using independent component analyses 4 and trial-based manual rejections. Average ERPs
Uncovering the neural correlates of inattentional deafness
The brain activity involved in processing auditory information has been extensively studied using Electroencephalography (EEG) techniques to measure Event-RelatedPotentials (ERPs), even in an aeronautical context [ 25 – 28 ]. The P300 component, one of the most commonly studied ERPs, reflects the detection of an expected but unpredictable target (the oddball) in a stream of stimuli [ 29 ]. It can be elicited by the “oddball” paradigm. The P300 is typically ob- served in a time window between 300 to 600 ms after the auditory stimulus onset and reflects the occurrence of cognitive and attentional processes (c.f. [ 29 ] and [ 30 ] for a detailed review). When attentional focus deviates from the target, the P300 amplitude significantly decreases [ 31 ]. This link indicates that the P300 component is an excellent candidate to determine whether an auditory stimulus has broken through the attentional barrier. Importantly, it is gen- erally accepted that a distinction can be made between two subcomponents of the P300, name- ly the novelty P3 and the target P3 (also called P3b, or “classical” P3). Novelty P3 is a large positive deflection with a fronto-central scalp distribution that is elicited by novel, non-target stimuli and that mainly reflects involuntary attention shifts to changes in the environment [ 32 , 33 ]. It is functionally related to another subcomponent called P3a, that seems to be more specifically related to deviant auditory non-target events [ 34 ]. In contrast, the P3b, has a more posterior-parietal scalp distribution and a somewhat longer latency than novelty P3 and P3a. The P3b has been regarded as a sign of processes of memory access that are evoked by evalua- tion of stimuli in tasks that require some form of action like a covert or overt response, ecolog- ically closer to a real alarm occurring in a cockpit [ 35 ].
When the waves are sufficiently time-locked with respect to the reference time (usually the time of stimulation), the average can be performed directly in the time-domain (“event-relatedpotentials”, ERPs, in EEG and “event-related fields”, ERFs, in MEG). For activity with higher time dispersion with respect to one wave period, the resulting variation of phase across trials can cause the waves to cancel out in the average signal. This is particularly relevant for high- frequency activity (above 20 Hz), where a small time delay can cause a large phase difference. This cancellation can be circumvented by averaging the power of the signal in the time-frequency or time-scale plane. Several methods have been introduced for evaluating the average increase of energy in given frequency band, whether time-locked (“evoked” energy) or not (“induced” energy) (e.g. ).
6. Kotchoubey B, Lang S, Bostanov V, Birbaumer N. Is there a Mind? Electrophysiology of Unconscious Patients. News Physiol Sci 2002;17:38-42
7. Duncan C C at al. Event-relatedpotentials in clinical research: Guidelines for eliciting, recording, and quantifying mismatch, P300, and N400.Clin Neurophysiol 2009;1883-908.
Stimuli were presented via earphones and data were acquired at the bedside. Event-relatedpotentials’ recordings were performed with the eyes closed and minimal ambient noise. Event- relatedpotentials were recorded from Fz, Cz and Pz (Klem, Luders, Jasper, & Elger, 1999) and ref- erenced to the nose. Electro-oculogram (EOG) was acquired using 2 electrodes placed diagonally above and below the right eye. A ground elec- trode was placed near Fz and impedances were kept below 5 kΩ. Data were collected at a sampling rate of 500 Hz using a NuAmp EEG amplifier (NeuroSoft, Sterling, VA, USA) with analog bandpass filtering of 0.1–200 Hz. Stand- ardized stimulation (i.e., deep pressure and audi- tory stimuli) was performed prior to each sequence to improve arousal (Giacino, Kalmar, & Whyte, 2004). Additionally, event-related poten- tials was performed when bispectral monitoring (Aspect Medical Systems, Newton, USA) showed
between a non-changing situation in a changing environment, and is therefore a description of some change in reality. By extension any state of anything is also considered an event . There have been on-going attempts to formalize events into an upper event ontology, which could be defined as an event-related but application- independent ontology. An example of upper event ontology is presented in  under Order-Sorted Logic. A portion of the Dolce upper ontology  could also be considered as an upper event ontology in its own right, as Dolce lists its two top concept types as endurant (an entity that persists over time) and perdurant (an entity that happens in time or changes over time), which are both state and time related. Perdurant has two subtypes, one of which is the type event itself.
Finally, the finiteness of the set of values that can be returned by a probabilistic assignment is already ensured by the syntax for enumerated probabilistic assignments and by PO (event/assign/pWD3) for predicate probabilistic assignments and their non-emptyness is ensured by the standard feasability POs.
Inadequacy of adapted POs. Unfortunately, as we deal with potentially infinite-state systems, POs 1–3 presented above are not anymore sufficient for proving that the probability of eventually executing a non-convergent event or reaching a deadlock state is 1. Indeed, although the probability of decreasing the variant is always strictly positive because of PO (model/pVar) and although the number of values that can be returned by a given probabilistic assignment is always finite, the combination of event weights and parameter choice can make this value infinitely small in some cases. In this case, it is well known that almost-certain reachability/convergence is not ensured. This problem is a direct consequence of the unboundedness of the weights of convergent events, which, by getting arbitrarily big, cause the probability of decreasing the variant to get arbitrarily small. Examples illustrating this fact are given in Appendix B.
In standard within-subject analyses of event-related fMRI data, two steps are usually performed separately: detection of brain activity and estimation of the hemodynamic response. Because these two steps are inherently linked, we adopt the so-called region-based Joint Detection-Estimation (JDE) framework that addresses this joint issue using a multivariate inference for detection and estimation. JDE is built by making use of a regional bilinear generative model of the BOLD response and constraining the parameter estimation by physiological priors using temporal and spatial information in a Markovian model. In contrast to previous works that use Markov Chain Monte Carlo (MCMC) techniques to sample the resulting intractable posterior distribution, we recast the JDE into a missing data framework and derive a Variational Expectation-Maximization (VEM) algorithm for its inference. A variational approximation is used to approximate the Markovian model in the unsupervised spatially adaptive JDE inference, which allows automatic fine-tuning of spatial regularization parameters. It provides a new algorithm that exhibits interesting properties in terms of estimation error and computational cost compared to the previously used MCMC-based approach. Experiments on artificial and real data show that VEM-JDE is robust to model mis-specification and provides computational gain while maintaining good performance in terms of activation detection and hemodynamic shape recovery.