Nonconvulsive Epileptic Seizure Detection in Scalp EEG Using Multiway Data Analysis.

Automatic detection of non-convulsive seizures: A reduced complexity approach

Nonconvulsive seizures (NCS) and nonconvulsive status epilepticus (NCSE) are frequently observed in human patients. Diagnosis of NCS and NCSE only can be achieved by the use of electroencephalography (EEG). Electroencephalographic monitoring is rare in veterinary medicine and consequently there is limited data on frequency of NCS and NCSE. Determine the prevalence of NCS and NCSE in dogs and cats with a history of cluster seizures. Twenty-six dogs and 12 cats. Retrospective study. Medical records of dogs and cats with cluster seizures were reviewed. Electroencephalography was performed in order to identify electrographic seizure activity after the apparent cessation of convulsive seizure activity. Nonconvulsive seizures were detected in 9 dogs and 2 cats out of the 38 patients (29%). Nonconvulsive status epilepticus was detected in 4 dogs and 2 cats (16%). Five patients had both NCS and NCSE. A decreased level of consciousness was evident in 6/11 patients with NCS, 3/6 also had NCSE. Mortality rate for patients with NCS (73%) and NCSE (67%) was much higher than that for patients with no seizure activity on EEG (27%). Prevalence of NCS and NCSE is high in dogs and cats with a history of cluster seizures. Nonconvulsive seizures and NCSE are difficult to detect clinically and are associated with higher in hospital mortality rates. Results indicate that prompt EEG monitoring should be performed in dogs and cats with cluster seizures.

Nonconvulsive seizures (NCS) and nonconvulsive status epilepticus (NCSE) are electrographic seizures (ESz) that are not associated with overt clinical seizure activity. NCS are distinct ESz, whereas NCSE has ongoing, continuous electrographic seizure activity. Both are common in critically ill patients admitted to hospital intensive care units (ICUs), and studies have shown that about 20% of ICU patients undergoing continuous electroencephalography (cEEG) monitoring will have NCS/NCSE. Although the treatment for convulsive SE is well established, there is no clear consensus for the treatment of NCS/NCSE. Antiepileptic drugs (AEDs), such as phenytoin (PHT) and fosphenytoin (fPHT), used in convulsive SE are also used to treat NCS/NCSE despite lack of data for their appropriateness for these conditions. Recent studies have shown that very aggressive treatment of NCSss/NCSE can lead to worse outcomes because the AEDs used can have significant adverse effects. Recently, several intravenous (IV) AEDs have become available for substitution therapy when their oral use is not possible. There are retrospective case reports and case series that suggest that these AEDs may be beneficial for treatment of NCS/NCSE. The Treatment of Recurrent Electrographic Nonconvulsive Seizures (TRENdS) Study will compare the efficacy and tolerability of fPHT and lacosamide in patients having NCS as noted by cEEG monitoring. The study is currently open to recruitment and has 13 sites in the United States. A total of 200 subjects will be randomized, 100 to each treatment arm.

ObjectiveIn Western settings, non‐convulsive status epilepticus (NCSE) and non‐convulsive seizures (NCSz) are associated with high mortality. In comatose patients, interictal epileptiform discharges (IEDs) identified on routine electroencephalogram (EEG) are predictive of NCSE/NCS. Little is known regarding the prevalence, causes, or outcomes of NCSE/NCSz in sub‐Saharan Africa (SSA). We sought to investigate the prevalence of IEDs and NCSE/NCSz at a single teaching institution in SSA.MethodsFrom October 3, 2017, to May 21, 2018, adult inpatients on the internal medicine service at Zambia's University Teaching Hospital (UTH) with a Glasgow Coma Score (GCS) of ≤10 were identified, excluding patients with mechanical ventilation or open head wounds. Signed consent by a proxy was required for enrollment and 30‐minute EEG. Chart abstractions provided coma duration, presence/absence of clinical seizures during/prior to admission, history of epilepsy, and presumed coma etiology. A structured neurological examination was completed. Patients were followed to discharge or death. Risk factors for IEDs were evaluated.ResultsOf 392 eligible patients, 250 had EEGs. EEGs were not completed on eligible patients due to death (74), improved GCS (37), transfer within UTH (25), or lack of proxy (6). NCSE occurred in 22 of 250 (8.8%), NCSz in 3 of 250 (1.2%), and IEDs in 46 of 250 (18.4%) patients. Of the 250, 197 (78.8%) died. No specific risk factors for IEDs were identified.SignificanceIf the association between IEDs and NCSE among monitored populations in developed settings holds true for SSA, a projected 17%‐21% of comatose African adults have NCSE. No clinical characteristics identified those at risk.

Use of pregabalin for nonconvulsive seizures and nonconvulsive status epilepticus

Postictal deficits are a well-recognized phenomenon, yet their mechanisms and predictors remain poorly understood. Distinguishing postictal deficits from recurring nonconvulsive seizures (NCSz) or nonconvulsive status epilepticus (NCSE), their main differential diagnosis, can be challenging. This study aimed to identify clinical and electrographic features predictive of ongoing or recurrent ictal activity and to explore biomarkers associated with the duration and severity of postictal deficits. We conducted a retrospective case-control study of patients with prolonged (>1 h) postictal deficits at a tertiary medical center between 2017 and 2022. Based on electroencephalographic (EEG) findings, patients were classified as having uncomplicated Todd's palsy or recurring NCSz. Clinical, imaging, and electrographic characteristics were compared between groups, and factors associated with postictal duration were analyzed. Among 112 patients, 86 (77%) had uncomplicated Todd's palsy and 26 (23%) had recurring NCSz. Patients with NCSz were younger (63 vs. 76 years, p = .022) and had longer deficits (2.5 vs. 1.5 days, p = .024). Sporadic epileptiform discharges (73% vs. 22%, p < .001) and rhythmic or periodic discharges on the ictal-interictal continuum (50% vs. 16%, p = .001) were significantly more frequent in the NCSE group, even when limited to short EEGs or the first 30 min of continuous EEG (96% vs. 40%, p < .001). Consequently, 2HELPS2B scores were higher in NCSE patients (2 vs. 1, p = .0065). In the uncomplicated Todd's palsy group, a preceding focal to bilateral tonic-clonic seizure (1.5 vs. 1.0 days, p = .04) and any rhythmic or periodic discharges on EEG (3.0 vs. 1.5 days, p = .003) were associated with longer deficit duration. A notable minority of patients with prolonged postictal deficits have recurring NCSz. In the absence of reliable clinical or imaging predictors, patients with prolonged deficits may benefit from a 30-min EEG followed-when 2HELPS2B score > 1-by extended continuous EEG monitoring.

Non-convulsive seizures (NCS) are common among critically ill children with acute encephalopathy. Continuous electroencephalogram (CEEG) monitoring is an indispensable tool to detect NCS, which is essential to guiding management and assessing prognosis. Risk factors for NCS are highest in pediatric intensive care unit (PICU) patients with altered mental status (AMS) and a recently witnessed clinical seizure, acute changes on neuroimaging, and/or interictal abnormalities on CEEG. Screening for at least 24 hours in at risk pediatric populations is ideal, but around half of NCS may be detected within the first hour. Rapid treatment of prolonged seizures or status epilepticus is critical, as higher seizure burdens have been associated with poorer outcomes in critically ill children. This review integrates current information on critically ill children with AMS and the use of CEEGs, risk factors for NCS, duration of CEEG monitoring, and how the detection of NCS impacts management and outcomes.

Accurate classification of brain activity from electroencephalogram signals is essential for diagnosing neurological disorders such as epilepsy. In this paper, we propose an explainable deep learning method for epileptic seizure detection. The proposed approach converts electroencephalogram signals into audio waveforms, which are then transformed into time–frequency representations using two distinct continuous wavelet transforms, i.e., the Morlet and the Mexican Hat. These wavelet-based spectrograms effectively capture both temporal and spectral characteristics of the electroencephalogram signal data and serve as inputs to a set of convolutional neural network models with the aim to detect seizure activity. To improve model transparency, the proposed method integrates three class activation mapping techniques aimed to visualize the salient regions in the wavelet images that influence each prediction. Experimental evaluation on a real-world dataset emphasizes the efficacy of wavelet-based preprocessing in electroencephalogram signal analysis in prompt epileptic seizure detection, showing an accuracy equal to 0.922.

Aims: Point-of-care electroencephalogram (POC-EEG) is an acute care bedside screening tool for the identification of nonconvulsive seizures (NCS) and nonconvulsive status epilepticus (NCSE). The objective of this narrative review is to describe the economic themes related to POC-EEG in the United States (US). Materials and methods: We examined peer-reviewed, published manuscripts on the economic findings of POC-EEG for bedside use in US hospitals, which included those found through targeted searches on PubMed and Google Scholar. Conference abstracts, gray literature offerings, frank advertisements, white papers, and studies conducted outside the US were excluded. Results: Twelve manuscripts were identified and reviewed; results were then grouped into four categories of economic evidence. First, POC-EEG usage was associated with clinical management amendments and antiseizure medication reductions. Second, POC-EEG was correlated with fewer unnecessary transfers to other facilities for monitoring and reduced hospital length of stay (LOS). Third, when identifying NCS or NCSE onsite, POC-EEG was associated with greater reimbursement in Medical Severity-Diagnosis Related Group coding. Fourth, POC-EEG may lower labor costs via decreasing after-hours requests to EEG technologists for conventional EEG (convEEG). Limitations: We conducted a narrative review, not a systematic review. The studies were observational and utilized one rapid circumferential headband system, which limited generalizability of the findings and indicated publication bias. Some sample sizes were small and hospital characteristics may not represent all US hospitals. POC-EEG studies in pediatric populations were also lacking. Ultimately, further research is justified. Conclusions: POC-EEG is a rapid screening tool for NCS and NCSE in critical care and emergency medicine with potential financial benefits through refining clinical management, reducing unnecessary patient transfers and hospital LOS, improving reimbursement, and mitigating burdens on healthcare staff and hospitals. Since POC-EEG has limitations (i.e. no video component and reduced montage), the studies asserted that it did not replace convEEG.

We present a new method for separation of the rhythms of the electroencephalogram (EEG) signal. The proposed method is based on the Hilbert-Huang transform (HHT). The HHT consists two steps namely empirical mode decomposition (EMD) and the Hilbert transform (HT). The EMD decomposes EEG signal into set of narrow-band intrinsic mode functions (IMFs), and the Hilbert transformation of these IMFs provide instantaneous frequency estimation of the IMFs. The instantaneous frequency estimation of IMFs have been used as a feature to identify the IMFs in order to separate rhythms of EEG signal. The central tendency measure (CTM) has been used to quantify the variability in second order difference (SOD) plots of rhythms of the EEG signal. The CTM parameter is very effective to discriminate epileptic seizure EEG signals from the seizure-free EEG signals. The experimental results show the effectiveness of the proposed method for epileptic seizure detection.

Electroencephalography in stroke Care: From acute monitoring to Post-Stroke epilepsy.

The prevalence of electrographic seizures or nonconvulsive status epilepticus and the effect of such seizures in children treated with extracorporeal cardiac life support are not known. We investigated the occurrence of electrographic abnormalities, including asymmetries in amplitude or frequency of the background rhythm and interictal activity in children undergoing extracorporeal cardiac life support and their association with seizures. We compared mortality and radiologic evidence of neurologic injury between patients with seizures and those without seizures. Retrospective review of medical records and the Extracorporeal Life Support Organization database. PICU at a single institution. All pediatric patients up to 18 years old, who had extracorporeal cardiac life support and continuous electroencephalography monitoring between the years 2006 and 2011. None. Nineteen patients treated with extracorporeal cardiac life support underwent continuous electroencephalography monitoring. Seizures occurred in four patients (21%) and were exclusively nonconvulsive in three patients. Two of these four patients had nonconvulsive status epilepticus. Interictal discharges on electroencephalography were associated with seizures (odds ratio, 19.5 [95% CI, 1.29-292.75]; p = 0.03). Only 50% of the seizures were detected in the first hour of monitoring, whereas all seizures were detected within 24 hours. All patients with seizures had structural abnormalities seen on neuroimaging. Seizures were not significantly associated with increased mortality. To evaluate for ascertainment bias, we compared outcomes between patients who underwent extracorporeal cardiac life support and received continuous electroencephalography monitoring and those patients who underwent extracorporeal cardiac life support during the study period but did not receive electroencephalography (n = 30). Seizures are common in children during extracorporeal cardiac life support, and most seizures are nonconvulsive. In patients undergoing extracorporeal cardiac life support, clinical features are unreliable indicators of the presence of seizures. The presence of seizures is suggestive of CNS injury. This study is limited by the exclusion of neonates, a feature of the clinical use of electroencephalography at our institution. Although seizures were not associated with increased mortality, further prospective studies in larger populations are needed to assess the long-term morbidity associated with seizures during extracorporeal cardiac life support.

Comparative analysis of ANN performance of four feature extraction methods used in the detection of epileptic seizures

The Canonical Polyadic (CP) decomposition of R-way arrays is a powerful tool in multilinear algebra. Algorithms to compute an approximate CP decomposition from noisy observations are often based on Alternating Least Squares (ALS) which may require a large number of iterations to converge. To avoid this drawback we investigate semi-algebraic approaches that algebraically reformulate the CP decomposition into a set of simultaneous matrix diagonalization (SMD) problems. In particular, we propose a SEmi-algebraic framework for approximate CP decompositions via SImultaneous matrix diagonalization (SMD) and generalized unfoldings (SECSI-GU). SECSI-GU combines the benefits of two existing semi-algebraic approaches based on SMDs: the SECSI framework which selects the model estimate from multiple candidates obtained by solving multiple SMDs and the “Semi-Algebraic Tensor Decomposition” (SALT) algorithm which considers a “generalized” unfolding of the tensor in order to enhance the identifiability for tensors with R > 3 dimensions. The resulting SECSI-GU framework offers a large number of degrees of freedom to flexibly adapt the performance-complexity trade-off. As we show in numerical simulations, it outperforms SECSI and SALT for tensors with R > 3 dimensions.

Acute anoxic injury is common in the adult population, occurring most commonly due to sudden cardiac arrest. In patients that remain comatose after resuscitation, there is a period of uncertainty, accompanied by high morbidity and mortality. Therapeutic hypothermia (TH) has been used for neuroprotection during this time period, though its true benefit remains uncertain. As a result, the focus shifts to determining prognosis, with EEG being one of the most widely used, well-studied resources available. From the time EEG became available, specific findings after acute anoxic injury have been detailed and correlated with outcomes. Multiple grading scales have been proposed to simplify clinical decision-making. As clinical protocols have evolved, and TH has become the standard of care, specific EEG patterns, including nonreactivity, low-voltage output pattern, and nonconvulsive seizures (NCS) and nonconvulsive status epilepticus (NCSE) remain worrisome. When combined with complementary clinical data, EEG remains a powerful tool for predicting prognosis after acute anoxic injury. In this chapter, the pathophysiology of anoxic brain injury and value of TH will be presented. Various EEG features that are associated with anoxia and TH and their prognostic value will also be discussed.