Taking the EEG Back Into the Brain: The Power of Multiple Discrete Sources

Magnetoencephalography (MEG) allows the non-invasive detection of interictal epileptiform discharges (IEDs). Clinical MEG analysis in epileptic patients traditionally relies on the visual identification of IEDs, which is time consuming and partially subjective. Automatic, data-driven detection methods exist but show limited performance. Still, the rise of deep learning (DL)—with its ability to reproduce human-like abilities—could revolutionize clinical MEG practice. Here, we developed and validated STIED, a simple yet powerful supervised DL algorithm combining two convolutional neural networks with temporal (1D time-course) and spatial (2D topography) features of MEG signals inspired from current clinical guidelines. Our DL model enabled a successful identification of IEDs in patients suffering from focal epilepsy with frequent and high amplitude spikes (FE group), with high-performance metrics—accuracy, specificity, and sensitivity all exceeding 85%—when learning from spatiotemporal features of IEDs. This performance can be attributed to our handling of input data, which mimics established clinical MEG practice. Reverse engineering further revealed that STIED encodes fine spatiotemporal features of IEDs rather than their mere amplitude. The model trained on the FE group also showed promising results when applied to a separate group of presurgical patients with different types of refractory focal epilepsy, though further work is needed to distinguish IEDs from physiological transients. This proof-of-concept study represents a first step towards the use of STIED and DL algorithms in the routine clinical MEG evaluation of epilepsy.

Electroencephalogram(EEG)signals are mainly generated by postsynaptic potentials(EPSP)in the apical dendrites of pyramidal cells in the cortex. The mechanism of generation of interictal epileptiform discharges(IED)is paroxysmal depolarization shift(PDS). IEDs are negative towards the cortical surface and positive towards the white matter. To interpret EEG, readers should know normal EEG waveforms of wakefulness and sleep, normal variants such as wicket spikes or small sharp spikes, and artifacts such as eye movements or electromyography. Overinterpretation of EEG is a major cause of misdiagnosis; therefore, the accurate identification of IED is essential. The International Federation of Clinical Neurophysiology has proposed a criteria for the identification of IED. Long-term video-EEG is useful for correct diagnosis and localization of the epileptogenic zone in the setting of presurgical evaluation of intractable epilepsy and is also useful for the differential diagnosis of epileptic seizures and non-epileptic events. Clinical history and symptomatology are of utmost importance for correct diagnosis. Diagnosis should not be made using only EEG findings.

Concurrent continuous EEG-fMRI is a novel multimodal technique that is finding its way into clinical practice in epilepsy. EEG timeseries are used to identify the timing of interictal epileptiform discharges (IEDs) which is then included in a GLM analysis in fMRI to localize the epileptic onset zone. Nevertheless, there are still some concerns about its reliability concerning BOLD changes correlated with IEDs. Even though IEDs are identified by an experienced neurologist-epiliptologist, the reliability and concordance of the mark-ups is depending on many factors including the level of fatigue, the amount of time that he spent or, in some cases, even the screen that is being used for the display of timeseries. This investigation is aiming to unravel the effect of misidentification or inaccuracy in the mark-ups of IEDs in the fMRI statistical parametric maps. Concurrent EEG-fMRI was conducted in six subjects with various types of epilepsy. IEDs were identified by an experienced neurologist-epiliptologist. Analysis of EEG was performed with EEGLAB and analysis of fMRI was conducted in FSL. Preliminary results revealed lower statistical significance for missing events or larger period of IEDs than the actual ones and the introduction of false positives and false negatives in statistical parametric maps when random events were included in the GLM on top of the IEDs. Our results suggest that mark-ups in EEG for simultaneous EEG-fMRI should be done with caution from an experienced and restful neurologist as it affects the fMRI results in various and unpredicted ways.

Scalp electroencephalography (EEG) and intraoperative electrocorticography (ECoG) are routinely used in the evaluation of magnetic resonance imaging-negative temporal lobe epilepsy (TLE) undergoing standard anterior temporal lobectomy with amygdalohippocampectomy (ATL), but the utility of interictal epileptiform discharge (IED) identification and its role in outcome are poorly defined. To determine whether the following are associated with surgical outcomes in patients with magnetic resonance imaging-negative TLE who underwent standard ATL: (1) unilateral-only IEDs on preoperative scalp EEG; (2) complete resection of tissue generating IEDs on ECoG; (3) complete resection of opioid-induced IEDs recorded on ECoG; and (4) location of IEDs recorded on ECoG. Data were gathered through retrospective medical record review at a tertiary referral center. Adult and pediatric patients with TLE who underwent standard ATL between January 1, 1990, and October 15, 2010, were considered for inclusion. Inclusion criteria were magnetic resonance imaging-negative TLE, standard ECoG performed at the time of surgery, and a minimum follow-up of 12 months. Univariate analysis was performed using log-rank time-to-event analysis. Variables reaching significance with log-rank testing were further analyzed using Cox proportional hazards. Excellent or nonexcellent outcome at time of last follow-up. An excellent outcome was defined as Engel class I and a nonexcellent outcome as Engel classes II through IV. Eighty-seven patients met inclusion criteria, with 48 (55%) achieving an excellent outcome following ATL. Unilateral IEDs on scalp EEG (P = .001) and complete resection of brain regions generating IEDs on baseline intraoperative ECoG (P = .02) were associated with excellent outcomes in univariate analysis. Both were associated with excellent outcomes when analyzed with Cox proportional hazards (unilateral-only IEDs, relative risk = 0.31 [95% CI, 0.16-0.64]; complete resection of IEDs on baseline ECoG, relative risk = 0.39 [95% CI, 0.20-0.76]). Overall, 25 of 35 patients (71%) with both unilateral-only IEDs and complete resection of baseline ECoG IEDs had an excellent outcome. Unilateral-only IEDs on preoperative scalp EEG and complete resection of IEDs on baseline ECoG are associated with better outcomes following standard ATL in magnetic resonance imaging-negative TLE. Prospective evaluation is needed to clarify the use of ECoG in tailoring temporal lobectomy.

Analysis of EEG–fMRI data in focal epilepsy based on automated spike classification and Signal Space Projection

Since it has become possible to acquire EEG und fMRI data simultaneously (EEG-correlated fMRI, EEG/fMRI), this technique has been applied mainly to four fields: 1) mapping of fMRI activation associated with interictal epileptiform discharges (IED) in patients with epilepsy, 2) mapping of fMRI activation associated with physiological EEG activity, like alpha rhythm; 3) monitoring vigilance and sleep stages during fMRI experiments; and 4) simultaneous acquisition of fMRI data and evoked potentials. Here, EEG/fMRI studies of IED and alpha- and beta-rhythm are presented. Several studies have shown that EEG-correlated fMRI is a practicable method to be applied to patients with epilepsy showing frequent IED on scalp EEG. However, the clinical interpretation of the fMRI maps remains difficult; mainly because EEG/fMRI results have not been systematically validated with a gold standard, which are intracranial recordings and outcome after epilepsy surgery. Currently, EEG-correlated fMRI has to be considered as a research tool providing insights to the pathophysiological processes underlying epileptic disorders and to the effects of epileptiform activity on cognition. It remains unclear if the method will also be used for the routine clinical work-up of epilepsy patients in the future. EEG/fMRI studies of alpha- and beta-rhythm in normal controls have shown that spontaneous fluctuations of the band power during rest are associated with fMRI signal changes, possibly representing alterations of attention-modulating networks (retrosplenial and fronto-parietal cortices). Similar fMRI activation patterns are found during generalized and focal IED in epileptic patients and might represent an fMRI signature of the cognitive effects of IED.

Objective. Interictal epileptiform discharges (IEDs) are an important and widely accepted biomarker used in the diagnosis of epilepsy based on scalp electroencephalography (EEG). Because the visual detection of IEDs has various limitations, including high time consumption and high subjectivity, a faster, more robust, and automated IED detector is strongly in demand. Approach. Based on deep learning, we proposed an end-to-end framework with multi-scale morphologic features in the time domain and correlation in sensor space to recognize IEDs from raw scalp EEG. Main Results. Based on a balanced dataset of 30 patients with epilepsy, the results of the five-fold (leave-6-patients-out) cross-validation shows that our model achieved state-of-the-art detection performance (accuracy: 0.951, precision: 0.973, sensitivity: 0.938, specificity: 0.968, F1 score: 0.954, AUC: 0.973). Furthermore, our model maintained excellent IED detection rates in an independent test on three datasets. Significance. The proposed model could be used to assist neurologists in clinical EEG interpretation of patients with epilepsy. Additionally, this approach combines multi-level output and correlation among EEG sensors and provides new ideas for epileptic biomarker detection in scalp EEG.

ObjectiveTo compare scalp-EEG recorded physiological ripples co-occurring with vertex waves to pathological ripples co-occurring with interictal epileptiform discharges (IEDs). MethodsWe marked ripples in sleep EEGs of children. We compared the start of ripples to vertex wave- or IED-start, and duration, frequency, and root mean square (RMS) amplitude of physiological and pathological ripples using multilevel modeling. Ripples were classified as physiological or pathological using linear discriminant analysis. ResultsWe included 40 children with and without epilepsy. Ripples started (χ2(1) = 38.59, p < 0.001) later if they co-occurred with vertex waves (108.2 ms after vertex wave-start) than if they co-occurred with IEDs (4.3 ms after IED-start). Physiological ripples had longer durations (75.7 ms vs 53.0 ms), lower frequencies (98.3 Hz vs 130.6 Hz), and lower RMS amplitudes (0.9 μV vs 1.8 μV, all p < 0.001) than pathological ripples. Ripples could be classified as physiological or pathological with 98 % accuracy. Ripples recorded in children with idiopathic or symptomatic epilepsy seemed to form two subgroups of pathological ripples. ConclusionsRipples co-occurring with vertex waves or IEDs have different characteristics and can be differentiated as physiological or pathological with high accuracy. SignificanceThis is the first study that compares physiological and pathological ripples recorded with scalp EEG.

Measuring neuro-haemodynamic correlates in the brain of epilepsy patients using EEG-fMRI has opened new avenues in clinical neuroscience, as these are two complementary methods for understanding brain function. In this study, we investigated three patients with drug-resistant reflex epilepsy using EEG-fMRI. Different types of reflex epilepsy such as eating, startle myoclonus, and hot water epilepsy were included in the study. The analysis of EEG-fMRI data was based on the visual identification of interictal epileptiform discharges on scalp EEG. The convolution of onset time and duration of these epilepsy spikes was estimated, and using these condition-specific effects in a general linear model approach, we evaluated activation of fMRI. Patients with startle myoclonus epilepsy experienced epilepsy in response to sudden sound or touch, in association with increased delta and theta activity with a spike-and-slow-wave pattern of interictal epileptiform discharges on EEG and fronto-parietal network activation pattern on SPECT and EEG-fMRI. Eating epilepsy was triggered by sight or smell of food and fronto-temporal discharges were noted on video-EEG (VEEG). Similarly, fronto-temporo-parietal involvement was noted on SPECT and EEG-fMRI. Hot water epilepsy was triggered by contact with hot water either in the bath or by hand immersion, and VEEG showed fronto-parietal involvement. SPECT and EEG fMRI revealed a similar fronto-parietal-occipital involvement. From these results, we conclude that continuous EEG recording can improve the modelling of BOLD changes related to interictal epileptic activity and this can thus be used to understand the neuro-haemodynamic substrates involved in reflex epilepsy.

Chapter 24 - Noninvasive Electroencephalography and Mesial Temporal Sclerosis

Clinical Neurophysiology: EEG–Video Monitoring

Chapter 2 - Focal epilepsies and focal disorders

Brief Potentially Ictal Rhythmic Discharges (BIRDs), initially described in neonates, have been shown to correlate with increased risk of seizures in both critically ill and non-critically ill adults. In critically ill patients, BIRDs are associated with acute brain injury and worse functional outcomes. In non-critically ill adults, BIRDs are seen in patients with epilepsy with a greater likelihood of having drug resistance. The location of BIRDs seems to better predict the seizure onset zone compared to other interictal epileptiform discharges. The definition of BIRDs includes Paroxysmal Fast Activity (PFA), and they have similar clinical significance regardless of the exact cut-off frequencies. Their potential as a biomarker for seizure activity and seizure onset zone has been suggested. In patients with status epilepticus, BIRDs also resolve or decrease when seizures resolve. Thus, if BIRDs are observed on scalp EEG, more extended EEG monitoring is recommended to estimate their seizure burden and to guide treatment. With the recent addition of BIRDs in the critical care EEG terminology, with future investigations, we may soon be able to reach a consensus about the definition of electrographic seizures and better understand their neurophysiology and clinical significance.

EEG–fMRI validation studies in comparison with icEEG: A review

Objective. In electroencephalography (EEG)-based brain–computer interface (BCI) systems for motor control tasks the conventional practice is to decode motor intentions by using scalp EEG. However, scalp EEG only reveals certain limited information about the complex tasks of movement with a higher degree of freedom. Therefore, our objective is to investigate the effectiveness of source-space EEG in extracting relevant features that discriminate arm movement in multiple directions. Approach. We have proposed a novel feature extraction algorithm based on supervised factor analysis that models the data from source-space EEG. To this end, we computed the features from the source dipoles confined to Brodmann areas of interest (BA4a, BA4p and BA6). Further, we embedded class-wise labels of multi-direction (multi-class) source-space EEG to an unsupervised factor analysis to make it into a supervised learning method. Main Results. Our approach provided an average decoding accuracy of 71% for the classification of hand movement in four orthogonal directions, that is significantly higher (>10%) than the classification accuracy obtained using state-of-the-art spatial pattern features in sensor space. Also, the group analysis on the spectral characteristics of source-space EEG indicates that the slow cortical potentials from a set of cortical source dipoles reveal discriminative information regarding the movement parameter, direction. Significance. This study presents evidence that low-frequency components in the source space play an important role in movement kinematics, and thus it may lead to new strategies for BCI-based neurorehabilitation.