Source analysis of epileptiform discharges in absence epilepsy using Magnetoencephalography (MEG)

Predicting the therapeutic response to valproic acid in childhood absence epilepsy through electroencephalogram analysis using machine learning

Childhood absence epilepsy (CAE) is a common type of idiopathic generalized epilepsy, manifesting as daily multiple absence seizures. Although seizures in most patients can be adequately controlled with first-line antiseizure medication (ASM), approximately 25 % of patients respond poorly to first-line ASM. In addition, an accurate method for predicting first-line medication responsiveness is lacking. We used the quantitative electroencephalogram (QEEG) features of patients with CAE along with machine learning to predict the therapeutic effects of valproic acid in this population. We enrolled 25 patients with CAE from multiple medical centers. Twelve patients who required additional medication for seizure control or who were shifted to another ASM and 13 patients who achieved seizure freedom with valproic acid within 6 months served as the nonresponder and responder groups. Using machine learning, we analyzed the interictal background EEG data without epileptiform discharge before ASM. The following features were analyzed: EEG frequency bands, Hjorth parameters, detrended fluctuation analysis, Higuchi fractal dimension, Lempel-Ziv complexity (LZC), Petrosian fractal dimension, and sample entropy (SE). We applied leave-one-out cross-validation with support vector machine, K-nearest neighbor (KNN), random forest, decision tree, Ada boost, and extreme gradient boosting, and we tested the performance of these models. The responders had significantly higher alpha band power and lower delta band power than the nonresponders. The Hjorth mobility, LZC, and SE values in the temporal, parietal, and occipital lobes were higher in the responders than in the nonresponders. Hjorth complexity was higher in the nonresponders than in the responders in almost all the brain regions, except for the leads FP1 and FP2. Using KNN classification with theta band power in the temporal lobe yielded optimal performance, with sensitivity of 92.31 %, specificity of 76.92 %, accuracy of 84.62 %, and area under the curve of 88.46 %.We used various EEG features along with machine learning to accurately predict whether patients with CAE would respond to valproic acid. Our method could provide valuable assistance for pediatric neurologists in selecting suitable ASM.

Neurostimulation for Generalized Epilepsy: Should Therapy be Syndrome-specific?

Current applications of neurostimulation for generalized epilepsy use a one-target-fits-all approach that is agnostic to the specific epilepsy syndrome and seizure type being treated. The authors describe similarities and differences between the 2 "archetypes" of generalized epilepsy-Lennox-Gastaut syndrome and Idiopathic Generalized Epilepsy-and review recent neuroimaging evidence for syndrome-specific brain networks underlying seizures. Implications for stimulation targeting and programming are discussed using 5 clinical questions: What epilepsy syndrome does the patient have? What brain networks are involved? What is the optimal stimulation target? What is the optimal stimulation paradigm? What is the plan for adjusting stimulation over time?

EEG Network Analysis in Epilepsy with Generalized Tonic-Clonic Seizures Alone

Many contradictory theories regarding epileptogenesis in idiopathic generalized epilepsy have been proposed. This study aims to define the network that takes part in the formation of the spike-wave discharges in patients with generalized tonic-clonic seizures alone (GTCSa) and elucidate the network characteristics. Furthermore, we intend to define the most influential brain areas and clarify the connectivity pattern among them. The data were collected from 23 patients with GTCSa utilizing low-density electroencephalogram (EEG). The source localization of generalized spike-wave discharges (GSWDs) was conducted using the Standardized low-resolution brain electromagnetic tomography (sLORETA) methodology. Cortical connectivity was calculated utilizing the imaginary part of coherence. The network characteristics were investigated through small-world propensity and the integrated value of influence (IVI). Source localization analysis estimated that most sources of GSWDs were in the superior frontal gyrus and anterior cingulate. Graph theory analysis revealed that epileptic sources created a network that tended to be regularized during generalized spike-wave activity. The IVI analysis concluded that the most influential nodes were the left insular gyrus and the left inferior parietal gyrus at 3 and 4 Hz, respectively. In conclusion, some nodes acted mainly as generators of GSWDs and others as influential ones across the whole network.

Temporal and spatial dynamic propagation of electroencephalogram by combining power spectral and synchronization in childhood absence epilepsy

Objective

During the transition from normal to seizure and then to termination, electroencephalography (EEG) signals have complex changes in time-frequency-spatial characteristics. The quantitative analysis of EEG characteristics and the exploration of their dynamic propagation in this paper would help to provide new biomarkers for distinguishing between pre-ictal and inter-ictal states and to better understand the seizure mechanisms.

Methods

Thirty-three children with absence epilepsy were investigated with EEG signals. Power spectral and synchronization were combined to provide the time-frequency-spatial characteristics of EEG and analyze the spatial distribution and propagation of EEG in the brain with topographic maps. To understand the mechanism of spatial-temporal evolution, we compared inter-ictal, pre-ictal, and ictal states in EEG power spectral and synchronization network and its rhythms in each frequency band.

Results

Power, frequency, and spatial synchronization are all enhanced during the absence seizures to jointly dominate the epilepsy process. We confirmed that a rapid diffusion at the onset accompanied by the frontal region predominance exists. The EEG power rapidly bursts in 2–4 Hz through the whole brain within a few seconds after the onset. This spatiotemporal evolution is associated with spatial diffusion and brain regions interaction, with a similar pattern, increasing first and then decreasing, in both the diffusion of the EEG power and the connectivity of the brain network during the childhood absence epilepsy (CAE) seizures. Compared with the inter-ictal group, we observed increases in power of delta and theta rhythms in the pre-ictal group (P < 0.05). Meanwhile, the synchronization of delta rhythm decreased while that of alpha rhythm enhanced.

Conclusion

The initiation and propagation of CAE seizures are related to the abnormal discharge diffusion and the synchronization network. During the seizures, brain activity is completely changed with the main component delta rhythm. Furthermore, this article demonstrated for the first time that alpha inhibition, which is consistent with the brain’s feedback regulation mechanism, is caused by the enhancement of the network connection. Temporal and spatial evolution of EEG is of great significance for the transmission mechanism, clinical diagnosis and automatic detection of absence epilepsy seizures.

Contributions of Magnetoencephalography to Understanding Mechanisms of Generalized Epilepsies: Blurring the Boundary Between Focal and Generalized Epilepsies?

According to the latest operational 2017 ILAE classification of epileptic seizures, the generalized epileptic seizure is still conceptualized as "originating at some point within and rapidly engaging, bilaterally distributed networks." In contrast, the focal epileptic seizure is defined as "originating within networks limited to one hemisphere." Hence, one of the main concepts of "generalized" and "focal" epilepsy comes from EEG descriptions before the era of source localization, and a presumed simultaneous bilateral onset and bi-synchrony of epileptiform discharges remains a hallmark for generalized seizures. Current literature on the pathophysiology of generalized epilepsy supports the concept of a cortical epileptogenic focus triggering rapidly generalized epileptic discharges involving intact corticothalamic and corticocortical networks, known as the cortical focus theory. Likewise, focal epilepsy with rich connectivity can give rise to generalized spike and wave discharges resulting from widespread bilateral synchronization. Therefore, making this key distinction between generalized and focal epilepsy may be challenging in some cases, and for the first time, a combined generalized and focal epilepsy is categorized in the 2017 ILAE classification. Nevertheless, treatment options, such as the choice of antiseizure medications or surgical treatment, are the reason behind the importance of accurate epilepsy classification. Over the past several decades, plentiful scientific research on the pathophysiology of generalized epilepsy has been conducted using non-invasive neuroimaging and postprocessing of the electromagnetic neural signal by measuring the spatiotemporal and interhemispheric latency of bi-synchronous or generalized epileptiform discharges as well as network analysis to identify diagnostic and prognostic biomarkers for accurate diagnosis of the two major types of epilepsy. Among all the advanced techniques, magnetoencephalography (MEG) and multiple other methods provide excellent temporal and spatial resolution, inherently suited to analyzing and visualizing the propagation of generalized EEG activities. This article aims to provide a comprehensive literature review of recent innovations in MEG methodology using source localization and network analysis techniques that contributed to the literature of idiopathic generalized epilepsy in terms of pathophysiology and clinical prognosis, thus further blurring the boundary between focal and generalized epilepsy.

Magnetoencephalography for epileptic focus localization based on Tucker decomposition with ripple window

AIMS

To improve the Magnetoencephalography (MEG) spatial localization precision of focal epileptic.

METHODS

306-channel simulated or real clinical MEG is estimated as a lower-dimensional tensor by Tucker decomposition based on Higher-order orthogonal iteration (HOOI) before the inverse problem using linearly constraint minimum variance (LCMV). For simulated MEG data, the proposed method is compared with dynamic imaging of coherent sources (DICS), multiple signal classification (MUSIC), and LCMV. For clinical real MEG of 31 epileptic patients, the ripples (80-250 Hz) were detected to compare the source location precision with spikes using the proposed method or the dipole-fitting method.

RESULTS

The experimental results showed that the positional accuracy of the proposed method was higher than that of LCMV, DICS, and MUSIC for simulation data. For clinical real MEG data, the positional accuracy of the proposed method was higher than that of dipole-fitting regardless of whether the time window was ripple window or spike window. Also, the positional accuracy of the ripple window was higher than that of the spike window regardless of whether the source location method was the proposed method or the dipole-fitting method. For both shallow and deep sources, the proposed method provided effective performance.

CONCLUSION

Tucker estimation of MEG for source imaging by ripple window is a promising approach toward the presurgical evaluation of epileptics.

The 10 Common Evidence-Supported Indications for MEG in Epilepsy Surgery: An Illustrated Compendium

Unfamiliarity with the indications for and benefits of magnetoencephalography (MEG) persists, even in the epilepsy community, and hinders its acceptance to clinical practice, despite the evidence. The wide treatment gap for patients with drug-resistant epilepsy and immense underutilization of epilepsy surgery had similar effects. Thus, educating referring physicians (epileptologists, neurologists, and neurosurgeons) both about the value of epilepsy surgery and about the potential benefits of MEG can achieve synergy and greatly improve the process of selecting surgical candidates. As a practical step toward a comprehensive educational process to benefit potential MEG users, current MEG referrers, and newcomers to MEG, the authors have elected to provide an illustrated guide to 10 everyday situations where MEG can help in the evaluation of people with drug-resistant epilepsy. They are as follows: (1) lacking or imprecise hypothesis regarding a seizure onset; (2) negative MRI with a mesial temporal onset suspected; (3) multiple lesions on MRI; (4) large lesion on MRI; (5) diagnostic or therapeutic reoperation; (6) ambiguous EEG findings suggestive of "bilateral" or "generalized" pattern; (7) intrasylvian onset suspected; (8) interhemispheric onset suspected; (9) insular onset suspected; and (10) negative (i.e., spikeless) EEG. Only their practical implementation and furtherance of personal and collective education will lead to the potentially impactful synergy of the two-MEG and epilepsy surgery. Thus, while fulfilling our mission as physicians, we must not forget that ignoring the wealth of evidence about the vast underutilization of epilepsy surgery - and about the usefulness and value of MEG in selecting surgical candidates - is far from benign neglect.

Changes of Ictal-Onset Epileptic Network Synchronicity in Childhood Absence Epilepsy: A Magnetoencephalography Study

Objective: To further understand the mechanisms underlying epileptic network and the characteristics of individual specific network, we conducted a study on brain network by magnetoencephalography (MEG) focusing on patients with childhood absence epilepsy (CAE). Methods: The network connectivity of 22 patients was investigated with MEG at the source level. Network connectivity of spikes and slow waves was computed with accumulated source imaging (ASI) and correlation analysis. Time-frequency analysis was used to characterize the network changes during the ictal-onset period of each patient and the potential factors. Results: We found that spectral power increased at around 1 s and distributed at 2-4 Hz in all patients. Ictal spikes simultaneously showed elevation of network connectivity, predominantly excitatory connections, when generalized firing activity spread to the overall brain. High-frequency oscillations (HFOs) were prone to detect overexcited neuronal firing in certain focal areas. Conclusions: Personal network changes during ictal onset had unique features in the time range and parallel seizure rhythm uniformly in every patient. There was an important time point for generalized discharges of the epileptic network. Ictal spiking activity played an important role in the epileptic network synchronicity of childhood absence epilepsy. Frequency oscillations provided references for locating abnormal changes in neuromagnetic signals.

EMS-Net: A Deep Learning Method for Autodetecting Epileptic Magnetoencephalography Spikes

Epilepsy is a neurological disorder characterized by sudden and unpredictable epileptic seizures, which incurs significant negative impacts on patients' physical, psychological and social health. A practical approach to assist with the clinical assessment and treatment planning for patients is to process magnetoencephalography (MEG) data to identify epileptogenic zones. As a widely accepted biomarker of epileptic foci, epileptic MEG spikes need to be precisely detected. Given that the visual inspection of spikes is time consuming, an automatic and efficient system with adequate accuracy for spike detection is valuable in clinical practice. However, current approaches for MEG spike autodetection are dependent on hand-engineered features. Here, we propose a novel multiview Epileptic MEG Spikes detection algorithm based on a deep learning Network (EMS-Net) to accurately and efficiently recognize the spike events from MEG raw data. The results of the leave-k-subject-out validation tests for multiple datasets (i.e., balanced and realistic datasets) showed that EMS-Net achieved state-of-the-art classification performance (i.e., accuracy: 91.82% - 99.89%; precision: 91.90% - 99.45%; sensitivity: 91.61% - 99.53%; specificity: 91.60% - 99.96%; f1 score: 91.70% - 99.48%; and area under the curve: 0.9688 - 0.9998).

Misleading Focal Clinical, Neurophysiologic, and Imaging Features in 2 Children With Generalized Epilepsy Who Underwent Invasive Electroencephalographic (EEG) Monitoring

Children and adults with genetic generalized epilepsy may have focal clinical seizure symptoms as well as electroencephalographic (EEG) findings. This may pose a diagnostic challenge to clinicians, especially when concomitant focal neuroimaging findings exist and the epilepsy is medically refractory. We sought to highlight the challenges that clinicians may face through the description of 2 children with suspected genetic generalized epilepsy who had both focal seizure symptoms and EEG/neuroimaging findings and underwent invasive EEG monitoring. Ultimately, invasive monitoring failed to demonstrate a focal origin for the seizures in both cases, and instead confirmed the presence of genetic generalized epilepsy. We demonstrate that ≥3-Hz generalized monomorphic spike and waves are less likely to represent secondary bilateral synchrony, that focal neuroimaging findings may not always be causal and that repeated hyperventilation is an essential activation procedure for genetic generalized epilepsy.

Source localization of epileptiform discharges in childhood absence epilepsy using a distributed source model: a standardized, low-resolution, brain electromagnetic tomography (sLORETA) study

Localizing the source of epileptiform discharges in generalized epilepsy has been controversial for the past few decades. Recent neuroimaging studies have shown that epileptiform discharges in generalized epilepsy can be localized to a particular region. Childhood absence epilepsy (CAE) is the most common generalized epilepsy in childhood and is considered the prototype of idiopathic generalized epilepsy (IGE). To better understand electrophysiological changes and their development in CAE, we investigated the origin of epileptiform discharges. We performed distributed source localization with standardized, low-resolution, brain electromagnetic tomography (sLORETA). In 16 children with CAE, sLORETA images corresponding to the midpoint of the ascending phase and the negative peak of the spike were obtained from a total of 242 EEG epochs (121 epochs at each timepoint). Maximal current source density (CSD) was mostly located in the frontal lobe (69.4%). At the gyral level, maximal CSD was most commonly in the superior frontal gyrus (39.3%) followed by the middle frontal gyrus (14.0%) and medial frontal gyrus (8.7%). At the hemisphere level, maximal CSD was dominant in the right cerebral hemisphere (63.6%). These results were consistent at the midpoint of the ascending phase and the negative peak of the spike. Our results demonstrated that the major source of epileptiform discharges in CAE was the frontal lobe. These results suggest that the frontal lobe is involved in generating CAE. This finding is consistent with recent studies that have suggested selective cortical involvement, especially in the frontal regions, in IGE.

Ictal Source Locations and Cortico-Thalamic Connectivity in Childhood Absence Epilepsy: Associations with Treatment Response

Childhood absence epilepsy (CAE), the most common pediatric epilepsy syndrome, is usually treated with valproic acid (VPA) and lamotrigine (LTG) in China. This study aimed to investigate the ictal source locations and functional connectivity (FC) networks between the cortices and thalamus that are related to treatment response. Magnetoencephalography (MEG) data from 25 patients with CAE were recorded at 300 Hz and analyzed in 1-30 Hz frequency bands. Neuromagnetic sources were volumetrically scanned with accumulated source imaging. The FC networks between the cortices and thalamus were evaluated at the source level through a connectivity analysis. Treatment outcome was assessed after 36-66 months following MEG recording. The children with CAE were divided into LTG responder, LTG non-responder, VPA responder and VPA non-responder groups. The ictal source locations and cortico-thalamic FC networks were compared to the treatment response. The ictal source locations in the post-dorsal medial frontal cortex (post-DMFC, including the medial primary motor cortex and the supplementary sensorimotor area) were observed in all LTG non-responders but in all LTG responders. At 1-7 Hz, patients with fronto-thalamo-parietal/occipital (F-T-P/O) networks were older than those with fronto-thalamic (F-T) networks or other cortico-thalamic networks (p = 0.000). The duration of seizures in patients with F-T-P/O networks at 1-7 Hz was longer than that in patients with F-T networks or other cortico-thalamic networks (p = 0.001). The ictal post-DMFC source localizations suggest that children with CAE might experience initial LTG monotherapy failure. Moreover, the cortico-thalamo-cortical network is associated with age. Finally, the cortico-thalamo-cortical network consists of anterior and posterior cortices and might contribute to the maintenance of discharges.