Exploring the potential of pretrained CNNs and time-frequency methods for accurate epileptic EEG classification: a comparative study.

Prompt diagnosis of epilepsy relies on accurate classification of automated electroencephalogram (EEG) signals. Several approaches have been developed to characterize epileptic EEG data; however, none of them have exploited time-frequency data to evaluate the effect of tweaking parameters in pretrained frameworks for EEG data classification. This study compares the performance of several pretrained convolutional neural networks (CNNs) namely, AlexNet, GoogLeNet, MobileNetV2, ResNet-18 and SqueezeNet for the localization of epilepsy EEG data using various time-frequency data representation algorithms. Continuous wavelet transform (CWT), empirical Fourier decomposition (EFD), empirical mode decomposition (EMD), empirical wavelet transform (EWT), and variational mode decomposition (VMD) were exploited for the acquisition of 2D scalograms from 1D data. The research evaluates the effect of multiple factors, including noisy versus denoised scalograms, different optimizers, learning rates, single versus dual channels, model size, and computational time consumption. The benchmark Bern-Barcelona EEG dataset is used for testing purpose. Results obtained show that the combination of MobileNetV2, Continuous Wavelet Transform (CWT) and Adam optimizer at a learning rate of 10-4, coupled with dual-data channels, provides the best performance metrics. Specifically, these parameters result in optimal sensitivity, specificity, f1-score, and classification accuracy, with respective values of 96.06%, 96.15%, 96.08%, and 96.10%. To further corroborate the efficacy of opted pretrained models on exploited Signal Decomposition (SD) algorithms, the classifiers are also being simulated on Temple University database at pinnacle modeling composition. A similar pattern in the outcome readily validate the findings of our study and robustness of deep learning models on epilepsy EEG scalograms.The conclusions drawn emphasize the potential of pretrained CNN-based models to create a robust, automated system for diagnosing epileptiform. Furthermore, the study offers insights into the effectiveness of varying time-frequency techniques and classifier parameters for classifying epileptic EEG data.

Epileptic Seizure Detection with Hybrid Time-Frequency EEG Input: A Deep Learning Approach.

The precise detection of epileptic seizure helps to prevent the serious consequences of seizures. As the electroencephalogram (EEG) reflects the brain activity of patients effectively, it has been widely used in epileptic seizure detection in the past decades. Recently, deep learning-based detection methods which automatically learn features from the EEG signals have attracted much attention. However, with deep learning-based detection methods, different input formats of EEG signals will lead to different detection performances. In this paper, we propose a deep learning-based epileptic seizure detection method with hybrid input formats of EEG signals, i.e., original EEG, Fourier transform of EEG, short-time Fourier transform of EEG, and wavelet transform of EEG. Convolutional neural networks (CNNs) are designed for extracting latent features from these inputs. A feature fusion mechanism is applied to integrate the learned features to generate a more stable syncretic feature for seizure detection. The experimental results show that our proposed hybrid method is effective to improve the seizure detection performance in few-shot scenarios.

A Deep Learning-Based Classification Method for Different Frequency EEG Data.

In recent years, the research on electroencephalography (EEG) has focused on the feature extraction of EEG signals. The development of convenient and simple EEG acquisition devices has produced a variety of EEG signal sources and the diversity of the EEG data. Thus, the adaptability of EEG classification methods has become significant. This study proposed a deep network model for autonomous learning and classification of EEG signals, which could self-adaptively classify EEG signals with different sampling frequencies and lengths. The artificial design feature extraction methods could not obtain stable classification results when analyzing EEG data with different sampling frequencies. However, the proposed depth network model showed considerably better universality and classification accuracy, particularly for EEG signals with short length, which was validated by two datasets.

Early childhood epilepsies: epidemiology, classification, aetiology, and socio-economic determinants.

Epilepsies of early childhood are frequently resistant to therapy and often associated with cognitive and behavioural comorbidity. Aetiology focused precision medicine, notably gene-based therapies, may prevent seizures and comorbidities. Epidemiological data utilizing modern diagnostic techniques including whole genome sequencing and neuroimaging can inform diagnostic strategies and therapeutic trials. We present a 3-year, multicentre prospective cohort study, involving all children under 3 years of age in Scotland presenting with epilepsies. We used two independent sources for case identification: clinical reporting and EEG record review. Capture-recapture methodology was then used to improve the accuracy of incidence estimates. Socio-demographic and clinical details were obtained at presentation, and 24 months later. Children were extensively investigated for aetiology. Whole genome sequencing was offered for all patients with drug-resistant epilepsy for whom no aetiology could yet be identified. Multivariate logistic regression modelling was used to determine associations between clinical features, aetiology, and outcome. Three hundred and ninety children were recruited over 3 years. The adjusted incidence of epilepsies presenting in the first 3 years of life was 239 per 100 000 live births [95% confidence interval (CI) 216-263]. There was a socio-economic gradient to incidence, with a significantly higher incidence in the most deprived quintile (301 per 100 000 live births, 95% CI 251-357) compared with the least deprived quintile (182 per 100 000 live births, 95% CI 139-233), χ2 odds ratio = 1.7 (95% CI 1.3-2.2). The relationship between deprivation and incidence was only observed in the group without identified aetiology, suggesting that populations living in higher deprivation areas have greater multifactorial risk for epilepsy. Aetiology was determined in 54% of children, and epilepsy syndrome was classified in 54%. Thirty-one per cent had an identified genetic cause for their epilepsy. We present novel data on the aetiological spectrum of the most commonly presenting epilepsies of early childhood. Twenty-four months after presentation, 36% of children had drug-resistant epilepsy (DRE), and 49% had global developmental delay (GDD). Identification of an aetiology was the strongest determinant of both DRE and GDD. Aetiology was determined in 82% of those with DRE, and 75% of those with GDD. In young children with epilepsy, genetic testing should be prioritized as it has the highest yield of any investigation and is most likely to inform precision therapy and prognosis. Epilepsies in early childhood are 30% more common than previously reported. Epilepsies of undetermined aetiology present more frequently in deprived communities. This likely reflects increased multifactorial risk within these populations.

Resting-state EEG theta activity reflects degree of genetic determination of the major epilepsy syndromes.

OBJECTIVE: To explore relationship between EEG theta activity and clinical data that imply the degree of genetic determination of epilepsy. METHODS: Clinical data of interest were epilepsy diagnosis and positive / negative family history of epilepsy. Study groups were: idiopathic generalized epilepsy (IGE), focal epilepsy (FE); FE of unknown etiology (FEUNK), FE of postnatal-acquired etiology (FEPA); all patients with positive / negative family history of epilepsy (FAPALL, FANALL, respectively), disregarding of the syndrome; FAP patients with 1st degree affected relative (FAP1) and those with 2nd degree epileptic relative only (FAP2). Quantitative EEG analysis assessed amount of theta (3.5-7.0 Hz) activity in 180 seconds of artifact-free waking EEG background activity for each patient and group. Group comparison was carried out by nonparametric statistics. RESULTS: Differences of theta activity were: FAPALL > FANALL (p = 0.01); FAP1 > FAP2 (p = 0.2752). IGE > FE (p = 0.02); FEUNK > FEPA (p = 0.07). CONCLUSIONS: This was the first attempt to explore and quantitatively ascertain relationship between an EEG variable and clinical data that imply greater or lesser degree of genetic determination in epilepsy. SIGNIFICANCE: Theta activity is endophenotype that bridges the gap between epilepsy susceptibility genes and clinical phenotypes. Amount of theta activity is indicative of degree of genetic determination of the epilepsies.

Sub-genic intolerance, ClinVar, and the epilepsies: A whole-exome sequencing study of 29,165 individuals.

Both mild and severe epilepsies are influenced by variants in the same genes, yet an explanation for the resulting phenotypic variation is unknown. As part of the ongoing Epi25 Collaboration, we performed a whole-exome sequencing analysis of 13,487 epilepsy-affected individuals and 15,678 control individuals. While prior Epi25 studies focused on gene-based collapsing analyses, we asked how the pattern of variation within genes differs by epilepsy type. Specifically, we compared the genetic architectures of severe developmental and epileptic encephalopathies (DEEs) and two generally less severe epilepsies, genetic generalized epilepsy and non-acquired focal epilepsy (NAFE). Our gene-based rare variant collapsing analysis used geographic ancestry-based clustering that included broader ancestries than previously possible and revealed novel associations. Using the missense intolerance ratio (MTR), we found that variants in DEE-affected individuals are in significantly more intolerant genic sub-regions than those in NAFE-affected individuals. Only previously reported pathogenic variants absent in available genomic datasets showed a significant burden in epilepsy-affected individuals compared with control individuals, and the ultra-rare pathogenic variants associated with DEE were located in more intolerant genic sub-regions than variants associated with non-DEE epilepsies. MTR filtering improved the yield of ultra-rare pathogenic variants in affected individuals compared with control individuals. Finally, analysis of variants in genes without a disease association revealed a significant burden of loss-of-function variants in the genes most intolerant to such variation, indicating additional epilepsy-risk genes yet to be discovered. Taken together, our study suggests that genic and sub-genic intolerance are critical characteristics for interpreting the effects of variation in genes that influence epilepsy.

The ILAE classification of seizures and the epilepsies: Modification for seizures in the neonate. Position paper by the ILAE Task Force on Neonatal Seizures.

Seizures are the most common neurological emergency in the neonatal period and in contrast to those in infancy and childhood, are often provoked seizures with an acute cause and may be electrographic-only. Hence, neonatal seizures may not fit easily into classification schemes for seizures and epilepsies primarily developed for older children and adults. A Neonatal Seizures Task Force was established by the International League Against Epilepsy (ILAE) to develop a modification of the 2017 ILAE Classification of Seizures and Epilepsies, relevant to neonates. The neonatal classification framework emphasizes the role of electroencephalography (EEG) in the diagnosis of seizures in the neonate and includes a classification of seizure types relevant to this age group. The seizure type is determined by the predominant clinical feature. Many neonatal seizures are electrographic-only with no evident clinical features; therefore, these are included in the proposed classification. Clinical events without an EEG correlate are not included. Because seizures in the neonatal period have been shown to have a focal onset, a division into focal and generalized is unnecessary. Seizures can have a motor (automatisms, clonic, epileptic spasms, myoclonic, tonic), non-motor (autonomic, behavior arrest), or sequential presentation. The classification allows the user to choose the level of detail when classifying seizures in this age group.

[Diagnostics in epilepsy - potential of magnetic resonance imaging]. / Diagnostik der Epilepsie ­ Möglichkeiten der Magnetresonanztomografie.

Epilepsy is a common neurologic disease frequently encountered by small animal practitioners. The disease comprises a multiplicity of clinical presentations and etiologies and often necessitates a comprehensive as well as cost-intensive diagnostic workup. This is mandatory in order to be able to diagnose or exclude a metabolic cause of the seizures and to distinguish between idiopathic and structural epilepsy. The examination by means of magnetic resonance imaging (MRI) represents a central component of the diagnostic workup, which in turn has essential effects on treatment and prognosis. In order to achieve standardized examination and comparable results, it is of utmost importance to use defined MRI protocols. Accordingly, communication and interaction between clinical institutions may be facilitated and as of yet undetected structural changes might be recorded in future MRI techniques. This review article sets particularly emphasis on the definition and classification of epilepsy as well as its diagnostic imaging procedures and refers to statistics and specialists' recommendations for the diagnostic workup in dogs.

Recent Advances in the Diagnosis and Treatment of Neonatal Seizures.

Seizures are the most common neurological emergency in the neonates, and this age group has the highest incidence of seizures compared with any other period of life. The author provides a narrative review of recent advances in the genetics of neonatal epilepsies, new neonatal seizure classification system, diagnostics, and treatment of neonatal seizures based on a comprehensive literature review (MEDLINE using PubMED and OvidSP vendors with appropriate keywords to incorporate recent evidence), personal practice, and experience. Knowledge regarding various systemic and postzygotic genetic mutations responsible for neonatal epilepsy has been exploded in recent times, as well as better delineation of clinical phenotypes associated with rare neonatal epilepsies. An International League Against Epilepsy task force on neonatal seizure has proposed a new neonatal seizure classification system and also evaluated the specificity of semiological features related to particular etiology. Although continuous video electroencephalogram (EEG) is the gold standard for monitoring neonatal seizures, amplitude-integrated EEGs have gained significant popularity in resource-limited settings. There is tremendous progress in the automated seizure detection algorithm, including the availability of a fully convolutional neural network using artificial machine learning (deep learning). There is a substantial need for ongoing research and clinical trials to understand optimal medication selection (first line, second line, and third line) for neonatal seizures, treatment duration of antiepileptic drugs after cessation of seizures, and strategies to improve neuromorbidities such as cerebral palsy, epilepsy, and developmental impairments. Although in recent times, levetiracetam use has been significantly increased for neonatal seizures, a multicenter, randomized, blinded, controlled phase IIb trial confirmed the superiority of phenobarbital over levetiracetam in the acute suppression of neonatal seizures. While there is no single best choice available for the management of neonatal seizures, institutional guidelines should be formed based on a consensus of local experts to mitigate wide variability in the treatment and to facilitate early diagnosis and treatment.

Epilepsy-Definition, Classification, Pathophysiology, and Epidemiology.

Seizures affect the lives of 10% of the global population and result in epilepsy in 1 to 2% of people around the world. Current knowledge about etiology, diagnosis, and treatments for epilepsy is constantly evolving. As more is learned, appropriate and updated definitions and classification systems for seizures and epilepsy are of the utmost importance. Without proper definitions and classification, many individuals will be improperly diagnosed and incorrectly treated. It is also essential for research purposes to have proper definitions, so that appropriate populations can be identified and studied. Imprecise definitions, failure to use accepted terminology, or inappropriate use of terminology hamper our ability to study and advance the field of epilepsy. This article begins by discussing the pathophysiology and epidemiology of epilepsy, and then covers the accepted contemporary definitions and classifications of seizures and epilepsies.

Basics of modern epilepsy classification and terminology.

BACKGROUND: The terminology and classification of seizures and epilepsy has undergone multiple revisions in the last several decades, which can lead to confusion and miscommunication amongst physicians and researchers. In 2017, the International League Against Epilepsy (ILAE) revised the classification of both seizures and epilepsy types in an effort to use less ambiguous terminology. Over time, definitions for status epilepticus, febrile seizures, and neonatal seizures have also evolved, as has the delineation of various epilepsy syndromes by age. METHODS: Review of the literature for old and new terminology and various epilepsy syndromes was accomplished using the PubMed database system. RESULTS: In the following article, we review old terminology for classifying seizures and epilepsy as compared to the new (2017) ILAE guidelines. We discuss neonatal seizures, status epilepticus, febrile seizures, autoimmune epilepsy and various epilepsy syndromes by age of onset. CONCLUSION: Adopting a classification system that uses plain language allows for more effective and efficient communication between individuals and across specialties. Definitions of various syndromes and seizure types have evolved over time and are reviewed.

Nocturnal events in children: When and how to evaluate.

Nocturnal events of wide variety and concern are frequently reported by patients and their caregivers. To evaluate suspected abnormal events, primary care physicians must first be familiar with normal behaviors, movements and breathing patterns. Abnormal nocturnal events can then be categorized as nocturnal seizure, parasomnia, sleep-related movement disorder or sleep-related breathing disorder. Diagnoses in the above categories can be made clinically; however, it is important to know when to refer for additional evaluation. Comprehensive literature review was undertaken of nocturnal and sleep-related disorders. This guide reviews nocturnal seizures, normal and abnormal nonepileptic movements and behaviors, discusses broad indications for referral for electroencephalography (EEG) or polysomnography (PSG), and guides counseling and management for patients and their families, ultimately aiding in interpretation of both findings and prognosis. Epilepsy syndromes can result in seizures during sleep or adjacent periods of wakefulness. Parasomnias and sleep-related movement disorders tend to also occur in childhood and may be distinguished clinically. Referral to additional specialists for specific studies including EEG or PSG can be necessary, while other times a knowledgeable and vigilant clinician can contribute to a prompt diagnosis based on clinical features. Nocturnal events often can be managed with parental reassurance and watchful waiting, but treatment or evaluation may be needed. Sleep-related breathing disorders are important to recognize as they present very differently in children than in adults and early intervention can be life-saving. This review should allow both primary and subspecialty non-neurologic pediatric and adolescent health care providers to better utilize EEG and PSG as part of a larger comprehensive clinical approach, distinguishing and managing both epileptic and nonepileptic nocturnal disorders of concern while fostering communication across providers to facilitate and coordinate better holistic long-term care of pediatric and adolescent patients.

SUDEP classification: Discordances between forensic investigators and epileptologists.

We compared sudden unexpected death in epilepsy (SUDEP) diagnosis rates between North American SUDEP Registry (NASR) epileptologists and original death investigators, to determine degree and causes of discordance. In 220 SUDEP cases with post-mortem examination, we recorded the epileptologist adjudications and medical examiner- and coroner- (ME/C) listed causes of death (CODs). COD diagnosis concordance decreased with NASR's uncertainty in the SUDEP diagnosis: highest for Definite SUDEP (84%, n = 158), lower in Definite Plus (50%, n = 36), and lowest in Possible (0%, n = 18). Rates of psychiatric comorbidity, substance abuse, and toxicology findings for drugs of abuse were all higher in discordant cases than concordant cases. Possible SUDEP cases, an understudied group, were significantly older, and had higher rates of cardiac, drug, or toxicology findings than more certain SUDEP cases. With a potentially contributing or competing COD, ME/Cs favored non-epilepsy-related diagnoses, suggesting a bias toward listing CODs with structural or toxicological findings; SUDEP has no pathognomonic features. A history of epilepsy should always be listed on death certificates and autopsy reports. Even without an alternate COD, ME/Cs infrequently classified COD as "SUDEP." Improved collaboration and communication between epilepsy and ME/C communities improve diagnostic accuracy, as well as bereavement and research opportunities.

Domino-like transient dynamics at seizure onset in epilepsy.

The International League Against Epilepsy (ILAE) groups seizures into "focal", "generalized" and "unknown" based on whether the seizure onset is confined to a brain region in one hemisphere, arises in several brain region simultaneously, or is not known, respectively. This separation fails to account for the rich diversity of clinically and experimentally observed spatiotemporal patterns of seizure onset and even less so for the properties of the brain networks generating them. We consider three different patterns of domino-like seizure onset in Idiopathic Generalized Epilepsy (IGE) and present a novel approach to classification of seizures. To understand how these patterns are generated on networks requires understanding of the relationship between intrinsic node dynamics and coupling between nodes in the presence of noise, which currently is unknown. We investigate this interplay here in the framework of domino-like recruitment across a network. In particular, we use a phenomenological model of seizure onset with heterogeneous coupling and node properties, and show that in combination they generate a range of domino-like onset patterns observed in the IGE seizures. We further explore the individual contribution of heterogeneous node dynamics and coupling by interpreting in-vitro experimental data in which the speed of onset can be chemically modulated. This work contributes to a better understanding of possible drivers for the spatiotemporal patterns observed at seizure onset and may ultimately contribute to a more personalized approach to classification of seizure types in clinical practice.

Interrater agreement of classification of photoparoxysmal electroencephalographic response.

Our goal was to assess the interrater agreement (IRA) of photoparoxysmal response (PPR) using the classification proposed by a task force of the International League Against Epilepsy (ILAE), and a simplified classification system proposed by our group. In addition, we evaluated IRA of epileptiform discharges (EDs) and the diagnostic significance of the electroencephalographic (EEG) abnormalities. We used EEG recordings from the European Reference Network (EpiCARE) and Standardized Computer-based Organized Reporting of EEG (SCORE). Six raters independently scored EEG recordings from 30 patients. We calculated the agreement coefficient (AC) for each feature. IRA of PPR using the classification proposed by the ILAE task force was only fair (AC = 0.38). This improved to a moderate agreement by using the simplified classification (AC = 0.56; P = .004). IRA of EDs was almost perfect (AC = 0.98), and IRA of scoring the diagnostic significance was moderate (AC = 0.51). Our results suggest that the simplified classification of the PPR is suitable for implementation in clinical practice.

An Automatic Epilepsy Detection Method Based on Improved Inductive Transfer Learning.

Epilepsy is a chronic disease caused by sudden abnormal discharge of brain neurons, causing transient brain dysfunction. The seizures of epilepsy have the characteristics of being sudden and repetitive, which has seriously endangered patients' health, cognition, etc. In the current condition, EEG plays a vital role in the diagnosis, judgment, and qualitative location of epilepsy among the clinical diagnosis of various epileptic seizures and is an indispensable means of detection. The study of the EEG signals of patients with epilepsy can provide a strong basis and useful information for in-depth understanding of its pathogenesis. Although, intelligent classification technologies based on machine learning have been widely used to the classification of epilepsy EEG signals and show the effectiveness. In fact, it is difficult to ensure that there is always enough EEG data available for training the model in real life, which will affect the performance of the algorithms. In view of this, to reduce the impact of insufficient data on the detection performance of the algorithms, a novel discriminate least squares regression- (DLSR-) based inductive transfer learning method was introduced which is on the basis of DLSR and the inductive transfer learning. And, it is applied to promote the adaptability and accuracy of the epilepsy EEG signal recognition. The proposed method inherits the advantages of DLSR; it can be more suitable for classification scenarios by expanding the interval between different classes. Meanwhile, it can simultaneously use the data of the target domain and the knowledge of the source domain, which is helpful for getting better performance. The results show that the improved method has more advantages in EEG signal recognition comparing to several other representative methods.

A taxonomy of seizure dynamotypes.

Seizures are a disruption of normal brain activity present across a vast range of species and conditions. We introduce an organizing principle that leads to the first objective Taxonomy of Seizure Dynamics (TSD) based on bifurcation theory. The 'dynamotype' of a seizure is the dynamic composition that defines its observable characteristics, including how it starts, evolves and ends. Analyzing over 2000 focal-onset seizures from multiple centers, we find evidence of all 16 dynamotypes predicted in TSD. We demonstrate that patients' dynamotypes evolve during their lifetime and display complex but systematic variations including hierarchy (certain types are more common), non-bijectivity (a patient may display multiple types) and pairing preference (multiple types may occur during one seizure). TSD provides a way to stratify patients in complement to present clinical classifications, a language to describe the most critical features of seizure dynamics, and a framework to guide future research focused on dynamical properties.

Epileptic seizures have been recognized for centuries. But it was only in the 1930s that it was realized that seizures are the result of out-of-control electrical activity in the brain. By placing electrodes on the scalp, doctors can identify when and where in the brain a seizure begins. But they cannot tell much about how the seizure behaves, that is, how it starts, stops or spreads to other areas. This makes it difficult to control and prevent seizures. It also helps explain why almost a third of patients with epilepsy continue to have seizures despite being on medication. Saggio, Crisp et al. have now approached this problem from a new angle using methods adapted from physics and engineering. In these fields, "dynamics research" has been used with great success to predict and control the behavior of complex systems like electrical power grids. Saggio, Crisp et al. reasoned that applying the same approach to the brain would reveal the dynamics of seizures and that such information could then be used to categorize seizures into groups with similar properties. This would in effect create for seizures what the periodic table is for the elements. Applying the dynamics research method to seizure data from more than a hundred patients from across the world revealed 16 types of seizure dynamics. These "dynamotypes" had distinct characteristics. Some were more common than others, and some tended to occur together. Individual patients showed different dynamotypes over time. By constructing a way to classify seizures based on the relationships between the dynamotypes, Saggio, Crisp et al. provide a new tool for clinicians and researchers studying epilepsy. Previous clinical tools have focused on the physical symptoms of a seizure (referred to as the phenotype) or its potential genetic causes (genotype). The current approach complements these tools by adding the dynamotype: how seizures start, spread and stop in the brain. This approach has the potential to lead to new branches of research and better understanding and treatment of seizures.