Seizure prediction - ready for a new era

A new era in electroencephalographic monitoring? Subscalp devices for ultra-long-term recordings

Inaccurate subjective seizure counting poses treatment and diagnostic challenges and thus suboptimal quality in epilepsy management. The limitations of existing hospital- and home-based monitoring solutions are motivating the development of minimally invasive, subscalp, implantable electroencephalography (EEG) systems with accompanying cloud-based software. This new generation of ultra-long-term brain monitoring systems is setting expectations for a sea change in the field of clinical epilepsy. From definitive diagnoses and reliable seizure logs to treatment optimization and presurgical seizure foci localization, the clinical need for continuous monitoring of brain electrophysiological activity in epilepsy patients is evident. This paper presents the converging solutions developed independently by researchers and organizations working at the forefront of next generation EEG monitoring. The immediate value of these devices is discussed as well as the potential drivers and hurdles to adoption. Additionally, this paper discusses what the expected value of ultra-long-term EEG data might be in the future with respect to alarms for especially focal seizures, seizure forecasting, and treatment personalization.

Machine learning and wearable devices of the future

Machine learning (ML) is increasingly recognized as a useful tool in healthcare applications, including epilepsy. One of the most important applications of ML in epilepsy is seizure detection and prediction, using wearable devices (WDs). However, not all currently available algorithms implemented in WDs are using ML. In this review, we summarize the state of the art of using WDs and ML in epilepsy, and we outline future development in these domains. There is published evidence for reliable detection of epileptic seizures using implanted electroencephalography (EEG) electrodes and wearable, non-EEG devices. Application of ML using the data recorded with WDs from a large number of patients could change radically the way we diagnose and manage patients with epilepsy.

Seizure forecasting and cyclic control of seizures

Epilepsy is a unique neurologic condition characterized by recurrent seizures, where causes, underlying biomarkers, triggers, and patterns differ across individuals. The unpredictability of seizures can heighten fear and anxiety in people with epilepsy, making it difficult to take part in day-to-day activities. Epilepsy researchers have prioritized developing seizure prediction algorithms to combat episodic seizures for decades, but the utility and effectiveness of prediction algorithms has not been investigated thoroughly in clinical settings. In contrast, seizure forecasts, which theoretically provide the probability of a seizure at any time (as opposed to predicting the next seizure occurrence), may be more feasible. Many advances have been made over the past decade in the field of seizure forecasting, including improvements in algorithms as a result of machine learning and exploration of non-EEG-based measures of seizure susceptibility, such as physiological biomarkers, behavioral changes, environmental drivers, and cyclic seizure patterns. For example, recent work investigating periodicities in individual seizure patterns has determined that more than 90% of people have circadian rhythms in their seizures, and many also experience multiday, weekly, or longer cycles. Other potential indicators of seizure susceptibility include stress levels, heart rate, and sleep quality, all of which have the potential to be captured noninvasively over long time scales. There are many possible applications of a seizure-forecasting device, including improving quality of life for people with epilepsy, guiding treatment plans and medication titration, optimizing presurgical monitoring, and focusing scientific research. To realize this potential, it is vital to better understand the user requirements of a seizure-forecasting device, continue to advance forecasting algorithms, and design clear guidelines for prospective clinical trials of seizure forecasting.

Autonomic nervous system changes detected with peripheral sensors in the setting of epileptic seizures

A better understanding of the early detection of seizures is highly desirable as identification of an impending seizure may afford improved treatments, such as antiepileptic drug chronotherapy, or timely warning to patients. While epileptic seizures are known to often manifest also with autonomic nervous system (ANS) changes, it is not clear whether ANS markers, if recorded from a wearable device, are also informative about an impending seizure with statistically significant sensitivity and specificity. Using statistical testing with seizure surrogate data and a unique dataset of continuously recorded multi-day wristband data including electrodermal activity (EDA), temperature (TEMP) and heart rate (HR) from 66 people with epilepsy (9.9 ± 5.8 years; 27 females; 161 seizures) we investigated differences between inter- and preictal periods in terms of mean, variance, and entropy of these signals. We found that signal mean and variance do not differentiate between inter- and preictal periods in a statistically meaningful way. EDA signal entropy was found to be increased prior to seizures in a small subset of patients. Findings may provide novel insights into the pathophysiology of epileptic seizures with respect to ANS function, and, while further validation and investigation of potential causes of the observed changes are needed, indicate that epilepsy-related state changes may be detectable using peripheral wearable devices. Detection of such changes with wearable devices may be more feasible for everyday monitoring than utilizing an electroencephalogram.

Neural Memory Networks for Seizure Type Classification

Classification of seizure type is a key step in the clinical process for evaluating an individual who presents with seizures. It determines the course of clinical diagnosis and treatment, and its impact stretches beyond the clinical domain to epilepsy research and the development of novel therapies. Automated identification of seizure type may facilitate understanding of the disease, and seizure detection and prediction have been the focus of recent research that has sought to exploit the benefits of machine learning and deep learning architectures. Nevertheless, there is not yet a definitive solution for automating the classification of seizure type, a task that must currently be performed by an expert epileptologist. Inspired by recent advances in neural memory networks (NMNs), we introduce a novel approach for the classification of seizure type using electrophysiological data. We first explore the performance of traditional deep learning techniques which use convolutional and recurrent neural networks, and enhance these architectures by using external memory modules with trainable neural plasticity. We show that our model achieves a state-of-the-art weighted F1 score of 0.945 for seizure type classification on the TUH EEG Seizure Corpus with the IBM TUSZ preprocessed data. This work highlights the potential of neural memory networks to support the field of epilepsy research, along with biomedical research and signal analysis more broadly.

Epileptic Seizure Detection for Imbalanced Datasets Using an Integrated Machine Learning Approach

Epileptic Seizure (Epilepsy) is a neurological disorder that occurs due to abnormal brain activities. Epilepsy affects patients' health and lead to life-threatening situations. Early prediction of epilepsy is highly effective to avoid seizures. Machine Learning algorithms have been used to classify epilepsy from Electroencephalograms (EEG) data. These algorithms exhibited reduced performance when classes are imbalanced. This work presents an integrated machine learning approach for epilepsy detection, which can effectively learn from imbalanced data. This approach utilizes Principal Component Analysis (PCA) at the first stage to extract both high- and low- variant Principal Components (PCs), which are empirically customized for imbalanced data classification. Conventionally, PCA is used for dimension reduction of a dataset leveraging PCs with high variances. In this paper, we propose a model to show that PCs associated with low variances can capture the implicit pattern of minor class of a dataset. The selected PCs are then fed into different machine learning classifiers to predict seizures. We performed experiments on the Epileptic Seizure Recognition dataset to evaluate our model. The experimental results show the robustness and effectiveness of the proposed model.

Signal quality and patient experience with wearable devices for epilepsy management

Noninvasive wearable devices have great potential to aid the management of epilepsy, but these devices must have robust signal quality, and patients must be willing to wear them for long periods of time. Automated machine learning classification of wearable biosensor signals requires quantitative measures of signal quality to automatically reject poor-quality or corrupt data segments. In this study, commercially available wearable sensors were placed on patients with epilepsy undergoing in-hospital or in-home electroencephalographic (EEG) monitoring, and healthy volunteers. Empatica E4 and Biovotion Everion were used to record accelerometry (ACC), photoplethysmography (PPG), and electrodermal activity (EDA). Byteflies Sensor Dots were used to record ACC and PPG, the Activinsights GENEActiv watch to record ACC, and Epitel Epilog to record EEG data. PPG and EDA signals were recorded for multiple days, then epochs of high-quality, marginal-quality, or poor-quality data were visually identified by reviewers, and reviewer annotations were compared to automated signal quality measures. For ACC, the ratio of spectral power from 0.8 to 5 Hz to broadband power was used to separate good-quality signals from noise. For EDA, the rate of amplitude change and prevalence of sharp peaks significantly differentiated between good-quality data and noise. Spectral entropy was used to assess PPG and showed significant differences between good-, marginal-, and poor-quality signals. EEG data were evaluated using methods to identify a spectral noise cutoff frequency. Patients were asked to rate the usability and comfort of each device in several categories. Patients showed a significant preference for the wrist-worn devices, and the Empatica E4 device was preferred most often. Current wearable devices can provide high-quality data and are acceptable for routine use, but continued development is needed to improve data quality, consistency, and management, as well as acceptability to patients.

Understanding the dynamics of biological and neural oscillator networks through exact mean-field reductions: a review

Many biological and neural systems can be seen as networks of interacting periodic processes. Importantly, their functionality, i.e., whether these networks can perform their function or not, depends on the emerging collective dynamics of the network. Synchrony of oscillations is one of the most prominent examples of such collective behavior and has been associated both with function and dysfunction. Understanding how network structure and interactions, as well as the microscopic properties of individual units, shape the emerging collective dynamics is critical to find factors that lead to malfunction. However, many biological systems such as the brain consist of a large number of dynamical units. Hence, their analysis has either relied on simplified heuristic models on a coarse scale, or the analysis comes at a huge computational cost. Here we review recently introduced approaches, known as the Ott-Antonsen and Watanabe-Strogatz reductions, allowing one to simplify the analysis by bridging small and large scales. Thus, reduced model equations are obtained that exactly describe the collective dynamics for each subpopulation in the oscillator network via few collective variables only. The resulting equations are next-generation models: Rather than being heuristic, they exactly link microscopic and macroscopic descriptions and therefore accurately capture microscopic properties of the underlying system. At the same time, they are sufficiently simple to analyze without great computational effort. In the last decade, these reduction methods have become instrumental in understanding how network structure and interactions shape the collective dynamics and the emergence of synchrony. We review this progress based on concrete examples and outline possible limitations. Finally, we discuss how linking the reduced models with experimental data can guide the way towards the development of new treatment approaches, for example, for neurological disease.

Deep Convolutional Neural Network-Based Epileptic Electroencephalogram (EEG) Signal Classification

Electroencephalogram (EEG) signals contain vital information on the electrical activities of the brain and are widely used to aid epilepsy analysis. A challenging element of epilepsy diagnosis, accurate classification of different epileptic states, is of particular interest and has been extensively investigated. A new deep learning-based classification methodology, namely epileptic EEG signal classification (EESC), is proposed in this paper. This methodology first transforms epileptic EEG signals to power spectrum density energy diagrams (PSDEDs), then applies deep convolutional neural networks (DCNNs) and transfer learning to automatically extract features from the PSDED, and finally classifies four categories of epileptic states (interictal, preictal duration to 30 min, preictal duration to 10 min, and seizure). It outperforms the existing epilepsy classification methods in terms of accuracy and efficiency. For instance, it achieves an average classification accuracy of over 90% in a case study with CHB-MIT epileptic EEG data.

Quantification of tRNA fragments by electrochemical direct detection in small volume biofluid samples

Elevated levels of transfer RNA (tRNA) fragments were recently identified in plasma samples from people with epilepsy in advance of a seizure, indicting a potential novel class of circulating biomarker. Current methods for detection and quantitation of tRNA fragments (tRFs) include northern blotting, RNA sequencing or custom Taqman-based PCR assays. The development of a simple, at home or clinic-based test, would benefit from a simple and reliable method to detect the tRFs using small volumes of biofluids. Here we describe an electrochemical direct detection method based on electrocatalytic platinum nanoparticles to detect 3 specific tRFs: 5'AlaTGC, 5'GlyGCC, and 5'GluCTC. Using synthetic tRF mimics we showed this system was linear over 9 orders of magnitude with sub-attomolar limits of detection. Specificity was tested using naturally occurring mismatched tRF mimics. Finally, we quantified tRF levels in patient plasma and showed that our detection system recapitulates results obtained by qPCR. We have designed a tRF detection system with high sensitivity and specificity capable of quantifying tRFs in low volumes of plasma using benchtop apparatus. This is an important step in the development of a point-of-care device for quantifying tRFs in whole blood.

Critical slowing down as a biomarker for seizure susceptibility

The human brain has the capacity to rapidly change state, and in epilepsy these state changes can be catastrophic, resulting in loss of consciousness, injury and even death. Theoretical interpretations considering the brain as a dynamical system suggest that prior to a seizure, recorded brain signals may exhibit critical slowing down, a warning signal preceding many critical transitions in dynamical systems. Using long-term intracranial electroencephalography (iEEG) recordings from fourteen patients with focal epilepsy, we monitored key signatures of critical slowing down prior to seizures. The metrics used to detect critical slowing down fluctuated over temporally long scales (hours to days), longer than would be detectable in standard clinical evaluation settings. Seizure risk was associated with a combination of these signals together with epileptiform discharges. These results provide strong validation of theoretical models and demonstrate that critical slowing down is a reliable indicator that could be used in seizure forecasting algorithms.

Critical slowing (associated with increased variance and autocorrelation) can precede critical state transitions. Here, the authors show critical slowing can be used as a marker in seizure forecasting algorithms.

Prevention of sudden unexpected death in epilepsy: current status and future perspectives

Introduction: Sudden unexpected death in epilepsy (SUDEP) affects about 1 in 1000 people with epilepsy, and even more in medically refractory epilepsy. As most people are between 20 and 40 years when dying suddenly, SUDEP leads to a considerable loss of potential life years. The most important risk factors are nocturnal and tonic-clonic seizures, underscoring that supervision and effective seizure control are key elements for SUDEP prevention. The question of whether specific antiepileptic drugs are linked to SUDEP is still controversially discussed. Knowledge and education about SUDEP among health-care professionals, patients, and relatives are of outstanding importance for preventive measures to be taken, but still poor and widely neglected.Areas covered: This article reviews epidemiology, pathophysiology, risk factors, assessment of individual SUDEP risk and available measures for SUDEP prevention. Literature search was done using Medline and Pubmed in October 2019.Expert opinion: Significant advances in the understanding of SUDEP were made in the last decade which allow testing of novel strategies to prevent SUDEP. Promising current strategies target neuronal mechanisms of brain stem dysfunction, cardiac susceptibility for fatal arrhythmias, and reliable detection of tonic-clonic seizures using mobile health technologies.Abbreviations: AED, antiepileptic drug; CBZ, carbamazepine; cLQTS, congenital long QT syndrome; EMU, epilepsy monitoring unit; FBTCS, focal to bilateral tonic-clonic seizures; GTCS, generalized tonic-clonic seizures; ICA, ictal central apnea; LTG, lamotrigine; PCCA, postconvulsive central apnea; PGES, postictal generalized EEG suppression; SRI, serotonin reuptake inhibitor; SUDEP, sudden unexpected death in epilepsy; TCS, tonic-clonic seizures.

Controversies on the network theory of epilepsy: Debates held during the ICTALS 2019 conference

Debates on six controversial topics on the network theory of epilepsy were held during two debate sessions, as part of the International Conference for Technology and Analysis of Seizures, 2019 (ICTALS 2019) convened at the University of Exeter, UK, September 2-5 2019. The debate topics were (1) From pathologic to physiologic: is the epileptic network part of an existing large-scale brain network? (2) Are micro scale recordings pertinent for defining the epileptic network? (3) From seconds to years: do we need all temporal scales to define an epileptic network? (4) Is it necessary to fully define the epileptic network to control it? (5) Is controlling seizures sufficient to control the epileptic network? (6) Does the epileptic network want to be controlled? This article, written by the organizing committee for the debate sessions and the debaters, summarizes the arguments presented during the debates on these six topics.

Heterogeneity of Preictal Dynamics in Human Epileptic Seizures

It is generally understood that there is a preictal phase in the development of a seizure and this precictal period is the basis for seizure prediction attempts. The focus of this study is the preictal global spatiotemporal dynamics and its intra-patient variability. We analyzed preictal broadband brain connectivity from human electrocorticography (ECoG) recordings of 185 seizures (which included 116 clinical seizures) collected from 12 patients. ECoG electrodes record from only a part of the cortex, leaving large regions of the brain unobserved. Brain connectivity was therefore estimated using the sparse-plus-latent-regularized precision matrix (SLRPM) method, which calculates connectivity from partial correlations of the conditional statistics of the observed regions given the unobserved latent regions. Brain connectivity was quantified using eigenvector centrality (EC), from which a degree of heterogeneity was calculated for the preictal periods of all seizures in each patient. Results from the SLRPM method are compared to those from the sparse-regularized precision matrix (SRPM) and correlation methods, which do not account for the unobserved inputs when estimating brain connectivity. The degree of heterogeneity estimated by the SLRPM method is higher than those estimated by the SRPM and correlation methods for the preictal periods in most patients. These results reveal substantial heterogeneity or desynchronization among brain areas in the preictal period of human epileptic seizures. Furthermore, the SLRPM method identifies more onset channels from the preictal active electrodes compared to the SRPM and correlation methods. Finally, the correlation between the degree of heterogeneity and seizure severity of patients for SLRPM and SRPM methods were lower than that obtained from the correlation method. These results support recent findings suggesting that inhibitory neurons can have anti-seizure effects by inducing variability or heterogeneity across seizures. Understanding how this variability is linked to seizure initiation may lead to better predictions and controlling therapies.

Data-driven separation and estimation of atrial dynamics in very high-dimensional electrocardiograms from epilepsy patients

Across biomedical areas, there is a significant unmet need for multimodal biomarkers that can improve prediction of abnormal events such as seizures, heart and asthma attacks or stroke. These markers may be multimodal and may include electrophysiological measures estimated from noninvasive, routinely collected clinical data, such as electroencephalograms (EEG) and electrocardiograms (ECG). In epilepsy, seizure detection and prediction from noninvasive data remains a difficult problem in need of novel approaches and markers. The inherent noise in high-dimensional EEG signals and artifact contamination often severely impacts the sensitivity and specificity of otherwise promising biomarkers. Long-term epilepsy clinical studies typically collect continuous ECG from which additional features may be estimated and combined with EEG measures to improve sensitivity to ictogenesis and seizure specificity. Prior work has focused on ventricular activity and features of the QRS complex, but atrial activity may also be modulated by seizure evolution. Given the high dimension of the ECG collected continuously over several days, an entirely data-driven approach is proposed, based on which ECG signals may be separated into ventricular and atrial contributions and studied separately. The relationship of atrial dynamics to seizure occurrence is assessed in a small number of pediatric epilepsy patients with high-dimensional ECG.

Predicting epileptic seizures using nonnegative matrix factorization

This paper presents a procedure for the patient-specific prediction of epileptic seizures. To this end, a combination of nonnegative matrix factorization (NMF) and smooth basis functions with robust regression is applied to power spectra of intracranial electroencephalographic (iEEG) signals. The resulting time and frequency components capture the dominant information from power spectra, while removing outliers and noise. This makes it possible to detect structure in preictal states, which is used for classification. Linear support vector machines (SVM) with L1 regularization are used to select and weigh the contributions from different number of not equally informative channels among patients. Due to class imbalance in data, synthetic minority over-sampling technique (SMOTE) is applied. The resulting method yields a computationally and conceptually simple, interpretable model of EEG signals of preictal and interictal states, which shows a good performance for the task of seizure prediction on two datasets (the EPILEPSIAE and on the public Epilepsyecosystem dataset).

Electrocardiographic changes associated with epilepsy beyond heart rate and their utilization in future seizure detection and forecasting methods

The ability to assess seizure risk may help provide timely warnings and more personalized treatment plans for people with epilepsy (PWE). ECG changes are commonly observed in epilepsy which make ECG a promising candidate to monitor seizure risk. Most ECG research in this domain has focused on heart rate-related changes. However, several studies have identified a range of other peri-ictal ECG parameter changes that may potentially prove useful for seizure detection and forecasting. Here, we offer a systematic review of ECG changes in epilepsy outside of heart rate. We performed the systematic literature review according to PRISMA guidelines using key words related to ECG, SUDEP and epilepsy. We identified and screened 502 abstracts, read 110 full papers, and included 24 papers in the final review. Our results suggest that PWE may be more prone to cardiac conduction abnormalities than healthy controls. During interictal periods, PWE were more likely to have abnormal QTc intervals, ST segment abnormalities, elevated T Waves, early repolarization (ER), increased P Wave dispersion and PR intervals when compared to controls. Apart from these baseline abnormalities, changes during the pre-ictal and ictal states have been reported, with arrhythmias, QTc prolongation and ST segment changes being the most common. A better understanding of these state-dependent changes may afford less-cumbersome and less-stigmatizing epilepsy monitoring tools in the future.

Oscillotherapeutics - Time-targeted interventions in epilepsy and beyond

Oscillatory brain activities support many physiological functions from motor control to cognition. Disruptions of the normal oscillatory brain activities are commonly observed in neurological and psychiatric disorders including epilepsy, Parkinson's disease, Alzheimer's disease, schizophrenia, anxiety/trauma-related disorders, major depressive disorders, and drug addiction. Therefore, these disorders can be considered as common oscillation defects despite having distinct behavioral manifestations and genetic causes. Recent technical advances of neuronal activity recording and analysis have allowed us to study the pathological oscillations of each disorder as a possible biomarker of symptoms. Furthermore, recent advances in brain stimulation technologies enable time- and space-targeted interventions of the pathological oscillations of both neurological disorders and psychiatric disorders as possible targets for regulating their symptoms.

Ensembling crowdsourced seizure prediction algorithms using long-term human intracranial EEG

Seizure prediction is feasible, but greater accuracy is needed to make seizure prediction clinically viable across a large group of patients. Recent work crowdsourced state-of-the-art prediction algorithms in a worldwide competition, yielding improvements in seizure prediction performance for patients whose seizures were previously found hard to anticipate. The aim of the current analysis was to explore potential performance improvements using an ensemble of the top competition algorithms. The results suggest that minor increments in performance may be possible; however, the outcomes of statistical testing limit the confidence in these increments. Our results suggest that for the specific algorithms, evaluation framework, and data considered here, incremental improvements are achievable but there may be upper bounds on machine learning-based seizure prediction performance for some patients whose seizures are challenging to predict. Other more tailored approaches that, for example, take into account a deeper understanding of preictal mechanisms, patient-specific sleep-wake rhythms, or novel measurement approaches, may still offer further gains for these types of patients.

Seizure prediction and intervention

Epilepsy treatment is challenging due to a lack of essential diagnostic tools, including methods for reliable seizure detection in the ambulatory setting, to assess seizure risk over time and to monitor treatment efficacy. This lack of objective diagnostics constitutes a significant barrier to better treatments, raises methodological concerns about the antiseizure medication evaluation process and, to patients, is a main issue contributing to the disease burden. Recent years have seen rapid progress towards better diagnostics that meet these needs of epilepsy patients and clinicians. Availability of comprehensive data and the rise of more powerful computational analysis methods have driven progress in this area. Here, we provide an overview on data- and theory-driven approaches aimed at identifying methods to reliably detect and forecast seizures as well as to monitor brain excitability and treatment efficacy in epilepsy. We provide a particular account on neural criticality, the hypothesis that cortical networks may be poised in a critical state at the boundary between different types of dynamics, and discuss its role in informing diagnostics to track cortex excitability and seizure risk in recent experiments. With the further expansion of digitalization in medicine, tele-medicine and long-term, ambulatory monitoring, these computationally based methods may gain more relevance in epilepsy in the future. This article is part of the special issue entitled 'New Epilepsy Therapies for the 21st Century - From Antiseizure Drugs to Prevention, Modification and Cure of Epilepsy'.

[Mobile seizure monitoring in epilepsy patients]

Wearables are receiving much attention from both epilepsy patients and treating physicians, for monitoring of seizure frequency and warning of seizures. They are also of interest for the detection of seizure-associated risks of patients, for differential diagnosis of rare seizure types and prediction of seizure-prone periods. Accelerometry, electromyography (EMG), heart rate and further autonomic parameters are recorded to capture clinical seizure manifestations. Currently, a clinical use to document nocturnal motor seizures is feasible. In this review the available devices, data on the performance in the documentation of seizures, current options for clinical use and developments in data analysis are presented and critically discussed.

Bispectrum and Recurrent Neural Networks: Improved Classification of Interictal and Preictal States

This work proposes a novel approach for the classification of interictal and preictal brain states based on bispectrum analysis and recurrent Long Short-Term Memory (LSTM) neural networks. Two features were first extracted from bilateral intracranial electroencephalography (iEEG) recordings of dogs with naturally occurring focal epilepsy. Single-layer LSTM networks were trained to classify 5-min long feature vectors as preictal or interictal. Classification performances were compared to previous work involving multilayer perceptron networks and higher-order spectral (HOS) features on the same dataset. The proposed LSTM network proved superior to the multilayer perceptron network and achieved an average classification accuracy of 86.29% on held-out data. Results imply the possibility of forecasting epileptic seizures using recurrent neural networks, with minimal feature extraction.

Evaluation of machine learning methods for seizure prediction in epilepsy

Epilepsy affects about 50 million people worldwide of which one third is refractory to medication. An automated and reliable system that warns of impending seizures would greatly improve patient’s quality of life by overcoming the uncertainty and helplessness due to the unpredicted events. Here we present new seizure prediction results including a performance comparison of different methods. The analysis is based on a new set of intracranial EEG data that has been recorded in our working group during presurgical evaluation. We applied two different methods for seizure prediction and evaluated their performance pseudoprospectively. The comparison of this evaluation with common statistical evaluation reveals possible reasons for overly optimistic estimations of the performance of seizure forecasting systems.

No evidence for critical slowing down prior to human epileptic seizures

There is an ongoing debate whether generic early warning signals for critical transitions exist that can be applied across diverse systems. The human epileptic brain is often considered as a prototypical system, given the devastating and, at times, even life-threatening nature of the extreme event epileptic seizure. More than three decades of international effort has successfully identified predictors of imminent seizures. However, the suitability of typically applied early warning indicators for critical slowing down, namely, variance and lag-1 autocorrelation, for indexing seizure susceptibility is still controversially discussed. Here, we investigated long-term, multichannel recordings of brain dynamics from 28 subjects with epilepsy. Using a surrogate-based evaluation procedure of sensitivity and specificity of time-resolved estimates of early warning indicators, we found no evidence for critical slowing down prior to 105 epileptic seizures.

Precursors of seizures due to specific spatial-temporal modifications of evolving large-scale epileptic brain networks

Knowing when, where, and how seizures are initiated in large-scale epileptic brain networks remains a widely unsolved problem. Seizure precursors – changes in brain dynamics predictive of an impending seizure – can now be identified well ahead of clinical manifestations, but either the seizure onset zone or remote brain areas are reported as network nodes from which seizure precursors emerge. We aimed to shed more light on the role of constituents of evolving epileptic networks that recurrently transit into and out of seizures. We constructed such networks from more than 3200 hours of continuous intracranial electroencephalograms recorded in 38 patients with medication refractory epilepsy. We succeeded in singling out predictive edges and predictive nodes. Their particular characteristics, namely edge weight respectively node centrality (a fundamental concept of network theory), from the pre-ictal periods of 78 out of 97 seizures differed significantly from the characteristics seen during inter-ictal periods. The vast majority of predictive nodes were connected by most of the predictive edges, but these nodes never played a central role in the evolving epileptic networks. Interestingly, predictive nodes were entirely associated with brain regions deemed unaffected by the focal epileptic process. We propose a network mechanism for a transition into the pre-seizure state, which puts into perspective the role of the seizure onset zone in this transition and highlights the necessity to reassess current concepts for seizure generation and seizure prevention.

Cognitive effort decreases beta, alpha, and theta coherence and ends afterdischarges in human brain

OBJECTIVE

Mental activation has been reported to modify the occurrence of epileptiform activity. We studied its effect on afterdischarges.

METHOD

In 15 patients with implanted electrodes we presented cognitive tasks when afterdischarges occurred. We developed a wavelet cross-coherence function to analyze the electrocorticography before and after the tasks and compared findings when cognitive tasks did or did not result in afterdischarge termination. Six patients returned for functional MRI (fMRI) testing, using similar tasks.

RESULTS

Cognitive tasks often could terminate afterdischarges when direct abortive stimulation could not. Wavelet cross-coherence analysis showed that, when afterdischarges stopped, there was decreased coherence throughout the brain in the 7.13-22.53 Hz frequency ranges (p values 0.008-0.034). This occurred a) regardless of whether an area activated on fMRI and b) regardless of whether there were afterdischarges in the area.

CONCLUSIONS

It is known that cognitive tasks can alter localized or network synchronization. Our results show that they can change activity throughout the brain. These changes in turn can terminate localized epileptiform activity.

SIGNIFICANCE

Cognitive tasks result in diffuse brain changes that can modify focal brain activity. Combined with a seizure detection device, cognitive activation might provide a non-invasive method of terminating or modifying seizures.

Identifying signal-dependent information about the preictal state: A comparison across ECoG, EEG and EKG using deep learning

Background

The inability to reliably assess seizure risk is a major burden for epilepsy patients and prevents developing better treatments. Recent advances have paved the way for increasingly accurate seizure preictal state detection algorithms, primarily using electrocorticography (ECoG). To develop seizure forecasting for broad clinical and ambulatory use, however, less complex and invasive modalities are needed. Algorithms using scalp electroencephalography (EEG) and electrocardiography (EKG) have also achieved better than chance performance. But it remains unknown how much preictal information is in ECoG versus modalities amenable to everyday use – such as EKG and single channel EEG - and how to optimally extract that preictal information for seizure prediction.

Methods

We apply deep learning - a powerful method to extract information from complex data - on a large epilepsy data set containing multi-day, simultaneous recordings of EKG, ECoG, and EEG, using a variety of feature sets. We use the relative performance of our algorithms to compare the preictal information contained in each modality.

Results

We find that single-channel EKG contains a comparable amount of preictal information as scalp EEG with up to 21 channels and that preictal information is best extracted not with standard heart rate measures, but from the power spectral density. We report that preictal information is not preferentially contained in EEG or ECoG channels within the seizure onset zone.

Conclusion

Collectively, these insights may help to devise future prospective, minimally invasive long-term epilepsy monitoring trials with single-channel EKG as a particularly promising modality.

Robust closed-loop control of spike-and-wave discharges in a thalamocortical computational model of absence epilepsy

In this paper, we investigate the abatement of spike-and-wave discharges in a thalamocortical model using a closed-loop brain stimulation method. We first explore the complex states and various transitions in the thalamocortical computational model of absence epilepsy by using bifurcation analysis. We demonstrate that the Hopf and double cycle bifurcations are the key dynamical mechanisms of the experimental observed bidirectional communications during absence seizures through top-down cortical excitation and thalamic feedforward inhibition. Then, we formulate the abatement of epileptic seizures to a closed-loop tracking control problem. Finally, we propose a neural network based sliding mode feedback control system to drive the dynamics of pathological cortical area to track the desired normal background activities. The control system is robust to uncertainties and disturbances, and its stability is guaranteed by Lyapunov stability theorem. Our results suggest that the seizure abatement can be modeled as a tracking control problem and solved by a robust closed-loop control method, which provides a promising brain stimulation strategy.

An update for epilepsy research and antiepileptic drug development: Toward precise circuit therapy

Epilepsy involves neuronal dysfunction at molecular, cellular, and circuit levels. The understanding of the mechanism of the epilepsies has advanced greatly in the last three decades, especially in terms of their cellular and molecular basis. However, despite the availability of ~30 anti-epileptic drugs (AEDs) with diverse molecular targets, there are still many challenges (e.g. drug resistance, side effects) in pharmacological treatment of epilepsies today. Because molecular mechanisms are integrated at the level of neuronal circuits, we suggest a shift in epilepsy treatment and research strategies from the "molecular" level to the "circuit" level. Recent technological advances have facilitated circuit mechanistic discovery at each level and have paved the way for many opportunities of novel therapeutic strategies and AED development toward precise circuit therapy.

Statistical Properties and Predictability of Extreme Epileptic Events

The use of extreme events theory for the analysis of spontaneous epileptic brain activity is a relevant multidisciplinary problem. It allows deeper understanding of pathological brain functioning and unraveling mechanisms underlying the epileptic seizure emergence along with its predictability. The latter is a desired goal in epileptology which might open the way for new therapies to control and prevent epileptic attacks. With this goal in mind, we applied the extreme event theory for studying statistical properties of electroencephalographic (EEG) recordings of WAG/Rij rats with genetic predisposition to absence epilepsy. Our approach allowed us to reveal extreme events inherent in this pathological spiking activity, highly pronounced in a particular frequency range. The return interval analysis showed that the epileptic seizures exhibit a highly-structural behavior during the active phase of the spiking activity. Obtained results evidenced a possibility for early (up to 7 s) prediction of epileptic seizures based on consideration of EEG statistical properties.

Elevation in plasma tRNA fragments precede seizures in human epilepsy

Transfer RNAs (tRNAs) are a major class of noncoding RNA. Stress-induced cleavage of tRNA is highly conserved and results in tRNA fragments. Here we find specific tRNA fragments in plasma are associated with epilepsy. Small RNA sequencing of plasma samples collected during video-EEG monitoring of focal epilepsy patients identified significant differences in three tRNA fragments (5', 5'AlaTGC, and 5'GluCTC) from controls. Levels of these tRNA fragments were higher in pre-seizure than post-seizure samples, suggesting they may serve as biomarkers of seizure risk in epilepsy patients. In vitro studies confirmed that production and extracellular release of tRNA fragments was lower after epileptiform-like activity in hippocampal neurons. We designed PCR-based assays to quantify tRNA fragments in a cohort of pre- and post-seizure plasma samples from focal epilepsy patients and healthy controls (n = 32/group). Receiver operating characteristic analysis indicated that tRNA fragments potently distinguished pre- from post-seizure patients (area under the curve of 0.8-0.95). Elevated tRNA fragments levels were not detected in patients with psychogenic non-epileptic seizures, and did not result from medication tapering. This study identifies a novel class of epilepsy biomarker and reveals the potential existence of prodromal molecular patterns in blood that could be used to predict seizure risk.

Wearable and Flexible Textile Electrodes for Biopotential Signal Monitoring: A review

Wearable electronics is a rapidly growing field that recently started to introduce successful commercial products into the consumer electronics market. Employment of biopotential signals in wearable systems as either biofeedbacks or control commands are expected to revolutionize many technologies including point of care health monitoring systems, rehabilitation devices, human–computer/machine interfaces (HCI/HMIs), and brain–computer interfaces (BCIs). Since electrodes are regarded as a decisive part of such products, they have been studied for almost a decade now, resulting in the emergence of textile electrodes. This study presents a systematic review of wearable textile electrodes in physiological signal monitoring, with discussions on the manufacturing of conductive textiles, metrics to assess their performance as electrodes, and an investigation of their application in the acquisition of critical biopotential signals for routine monitoring, assessment, and exploitation of cardiac (electrocardiography, ECG), neural (electroencephalography, EEG), muscular (electromyography, EMG), and ocular (electrooculography, EOG) functions.

Traceability and dynamical resistance of precursor of extreme events

Extreme events occur in a variety of natural, technical, and societal systems and often have catastrophic consequences. Their low-probability, high-impact nature has recently triggered research into improving our understanding of generating mechanisms, providing early warnings as well as developing control strategies. For the latter to be effective, knowledge about dynamical resistance of a system prior to an extreme event is of utmost importance. Here we introduce a novel time-series-based and non-perturbative approach to efficiently monitor dynamical resistance and apply it to high-resolution observations of brain activities from 43 subjects with uncontrollable epileptic seizures. We gain surprising insights into pre-seizure dynamical resistance of brains that also provide important clues for success or failure of measures for seizure prevention. The novel resistance monitoring perspective advances our understanding of precursor dynamics in complex spatio-temporal systems with potential applications in refining control strategies.

Epileptic Seizure Suppression by Focal Brain Cooling With Recirculating Coolant Cooling System: Modeling and Simulation

A focal brain cooling system for treatment of refractory epilepsy that is implantable and wearable may permit patients with this condition to lead normal daily lives. We have developed such a system for cooling of the epileptic focus by delivery of cold saline to a cooling device that is implanted cranially. The outflow is pumped for circulation and cooled by a Peltier device. Here, we describe the design of the system and evaluate its feasibility by simulation. Mathematical models were constructed based on equations of fluid dynamics and data from a cat model. Computational fluid dynamics simulations gave the following results: 1) a cooling device with a complex channel structure gives a more uniform temperature in the brain; 2) a cooling period of <10 min is required to reach an average temperature of 25.0°Cat 2 mm below the brain surface, which is the target temperature for seizure suppression. This time is short enough for cooling of the brain before seizure onset after seizure prediction by an intracranial electroencephalogram-based algorithm; and 3) battery charging would be required once every several days for most patients. These results suggest that the focal brain cooling system may be clinically applicable.

An overview of deep learning in medical imaging focusing on MRI

What has happened in machine learning lately, and what does it mean for the future of medical image analysis? Machine learning has witnessed a tremendous amount of attention over the last few years. The current boom started around 2009 when so-called deep artificial neural networks began outperforming other established models on a number of important benchmarks. Deep neural networks are now the state-of-the-art machine learning models across a variety of areas, from image analysis to natural language processing, and widely deployed in academia and industry. These developments have a huge potential for medical imaging technology, medical data analysis, medical diagnostics and healthcare in general, slowly being realized. We provide a short overview of recent advances and some associated challenges in machine learning applied to medical image processing and image analysis. As this has become a very broad and fast expanding field we will not survey the entire landscape of applications, but put particular focus on deep learning in MRI. Our aim is threefold: (i) give a brief introduction to deep learning with pointers to core references; (ii) indicate how deep learning has been applied to the entire MRI processing chain, from acquisition to image retrieval, from segmentation to disease prediction; (iii) provide a starting point for people interested in experimenting and perhaps contributing to the field of deep learning for medical imaging by pointing out good educational resources, state-of-the-art open-source code, and interesting sources of data and problems related medical imaging.

Fractal and Multifractal Properties of Electrographic Recordings of Human Brain Activity: Toward Its Use as a Signal Feature for Machine Learning in Clinical Applications

The quantification of brain dynamics is essential to its understanding. However, the brain is a system operating on multiple time scales, and characterization of dynamics across time scales remains a challenge. One framework to study such dynamics is that of fractal geometry; and currently there exist several methods for the study of brain dynamics using fractal geometry. We aim to highlight some of the practical challenges of applying fractal geometry to brain dynamics—and as a putative feature for machine learning applications, and propose solutions to enable its wider use in neuroscience. Using intracranially recorded electroencephalogram (EEG) and simulated data, we compared monofractal and multifractal methods with regards to their sensitivity to signal variance. We found that both monofractal and multifractal properties correlate closely with signal variance, thus not being a useful feature of the signal. However, after applying an epoch-wise standardization procedure to the signal, we found that multifractal measures could offer non-redundant information compared to signal variance, power (in different frequency bands) and other established EEG signal measures. We also compared different multifractal estimation methods to each other in terms of reliability, and we found that the Chhabra-Jensen algorithm performed best. Finally, we investigated the impact of sampling frequency and epoch length on the estimation of multifractal properties. Using epileptic seizures as an example event in the EEG, we show that there may be an optimal time scale (i.e., combination of sampling frequency and epoch length) for detecting temporal changes in multifractal properties around seizures. The practical issues we highlighted and our suggested solutions should help in developing robust methods for the application of fractal geometry in EEG signals. Our analyses and observations also aid the theoretical understanding of the multifractal properties of the brain and might provide grounds for new discoveries in the study of brain signals. These could be crucial for the understanding of neurological function and for the developments of new treatments.