Extracting seizure onset from surface EEG with independent component analysis: Insights from simultaneous scalp and intracerebral EEG

Barborică, Andrei; Mı̂ndruță, Ioana; Sheybani, Laurent; Spinelli, Laurent; Oane, Irina; Pistol, Constantin; Donos, Cristian; López‐Madrona, Víctor J.; Vulliémoz, Serge; Bénar, Christian‐George

doi:10.1016/j.nicl.2021.102838

Cited by 18 publications

(18 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Overlapping epochs can be used in such a scenario to further reduce the chance of missing a seizure since there will be more seizure epochs. The present system can be further improved by pre-processing the data to remove artifacts using combined Blind source separation and independent component analysis [74], which has proven more useful for separating linearly mixed independent sources in EEG signals, including artifacts [75,76]. The influence of such pre-processing can improve the presented results and can be investigated in future studies.…”

Section: Discussionmentioning

confidence: 95%

Neuromorphic Architecture Accelerated Automated Seizure Detection in Multi-Channel Scalp EEG

Ambati

Raja

Alhameed

et al. 2022

Sensors

View full text Add to dashboard Cite

Epileptic focal seizures can be localized in the brain using tracer injections during or immediately after the incidence of a seizure. A real-time automated seizure detection system with minimal latency can help time the injection properly to find the seizure origin accurately. Reliable real-time seizure detection systems have not been clinically reported yet. We developed an anomaly detection-based automated seizure detection system, using scalp-electroencephalogram (EEG) data, which can be trained using a few seizure sessions, and implemented it on commercially available hardware with parallel, neuromorphic architecture—the NeuroStack. We extracted nonlinear, statistical, and discrete wavelet decomposition features, and we developed a graphical user interface and traditional feature selection methods to select the most discriminative features. We investigated Reduced Coulomb Energy (RCE) networks and K-Nearest Neighbors (k-NN) for its several advantages, such as fast learning no local minima problem. We obtained a maximum sensitivity of 91.14%±1.77% and a specificity of 98.77%±0.57% with 5 s epoch duration. The system’s latency was 12 s, which is within most seizure event windows, which last for an average duration of 60 s. Our results showed that the CD feature consumes large computation resources and excluding it can reduce the latency to 3.6 s but at the cost of lower performance 80% sensitivity and 97% specificity. We demonstrated that the proposed methodology achieves a high specificity and an acceptable sensitivity within a short delay. Our results indicated also that individual-based RCE are superior to population-based RCE. The proposed RCE networks has been compared to SVM and ANN as a baseline for comparison as they are the most common machine learning seizure detection methods. SVM and ANN-based systems were trained on the same data as RCE and K-NN with features optimized specifically for them. RCE nets are superior to SVM and ANN. The proposed model also achieves comparable performance to the state-of-the-art deep learning techniques while not requiring a sizeable database, which is often expensive to build. These numbers indicate that the system is viable as a trigger mechanism for tracer injection.

show abstract

Section: Discussionmentioning

confidence: 95%

Neuromorphic Architecture Accelerated Automated Seizure Detection in Multi-Channel Scalp EEG

Ambati

Raja

Alhameed

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…Comparison of ESI with stereo-EEG using such recordings found that the equivalent current dipole localization was on average 47.2 ± 23.2 mm away from the stereo-EEG seizure onset. 39 While in another study using dipole and maximum distributed source, the median distance error to the centroid of seizure onset electrodes was 30–33 mm. 40 However, validation of these studies is limited by subjective EEG seizure interpretation and restricted spatial sampling of stereo-EEG electrodes.…”

Section: Introductionmentioning

confidence: 94%

Validating EEG source imaging using intracranial electrical stimulation

Unnwongse

Rampp

Wehner

et al. 2022

Brain Communications

View full text Add to dashboard Cite

Electrical source imaging is used in the presurgical epilepsy evaluation and in cognitive neurosciences to localize neuronal sources of brain potentials recorded on EEG. This study evaluates spatial accuracy of electrical source imaging for known sources, using electrical stimulation potentials recorded on simultaneous stereo-EEG and 37-electrode scalp EEG, and identifies factors determining the localization error. In 11 patients undergoing simultaneous stereo-EEG and 37-electrode scalp EEG recordings, sequential series of 99-110 biphasic pulses (2-millisecond pulse width) were applied by bipolar electrical stimulation on adjacent contacts of implanted stereo-EEG electrodes. The scalp EEG correlates of stimulation potentials were recorded with a sampling rate of 30 kHz. Electrical source imaging of averaged stimulation potentials was calculated utilizing a dipole source model of peak stimulation potentials based on individual four compartment finite element method head models with various skull conductivities (range from 0.0413 to 0.001 S/m). Fitted dipoles with goodness of fit of ≥80% were included into the analysis. The localization error was calculated using Euclidean distance between the estimated dipoles and the center point of adjacent stimulating contacts. A total of 3,619 stimulation locations, respectively dipole localizations, were included in the evaluation. Mean localization errors ranged from 10.3 to 26 mm, depending on source depth and selected skull conductivity. The mean localization error increased with an increase of source depth (r (3,617) = [0.19], P = 0.000) and decreased with an increase of skull conductivity (r (3,617) = [-0.26], P = 0.000). High skull conductivities (0.0413-0.0118 S/m) yielded significantly lower localization errors for all source depths. For superficial sources (<20 mm from inner skull), all skull conductivities yielded insignificantly different localization error. However, for deeper sources, in particular > 40 mm, high skull conductivities of 0.0413 and 0.0206 S/m yielded significantly lower localization errors. In relation to stimulation locations, the majority of estimated dipoles moved outward-forward-downward to inward-forward-downward with a decrease of source depth and an increase of skull conductivity. Multivariate analysis revealed that an increase of source depth, number of skull holes, and white matter volume, while decrease of skull conductivity independently led to higher localization error. This evaluation of electrical source imaging accuracy using artificial patterns with high signal to noise ratio supports its application in presurgical epilepsy evaluation and cognitive neurosciences. In our artificial potentials model, optimizing the selected skull conductivity minimized the localization error. Future studies should examine if this accounts for true neural signals.

show abstract

“…The interictal cerebral FDG-PET scan revealed extensive right temporal lobe hypometabolism: temporal pole, mesial and basal areas, and superior temporal and Heschl gyri ( Figure 1 ). To better characterize the electrical seizure onset and the early propagation network, we performed independent component analysis (ICA) of scalp EEG signals ( Figure 2A ) ( 10 , 11 ) using the infomax algorithm ( 12 ) as implemented in EEGLAB ( 13 ). One independent component (IC) was visually selected by an expert epileptologist based on the IC activation pattern ( Figure 2B ) and its time–frequency decomposition ( Figure 2C ).…”

Section: Casementioning

confidence: 99%

“…The treatment of musicogenic seizures usually ranges from the avoidance of the musical triggers together with anti-seizure medication to epilepsy surgery (2). As there is emerging evidence of an association between musicogenic seizures and anti-glutamic acid decarboxylase antibodies (anti-GAD abs) encephalitis, the focus progressively shifts toward immunotherapy (8,9). This article aimed to present two cases of musicogenic epilepsy of different etiologies (FCD type IIA and anti-GAD abs encephalitis) that involve the left or right temporal lobe, with their particularities of diagnosis and management of subsequent learning points.…”

Section: Introductionmentioning

confidence: 99%

Musicogenic seizures in temporal lobe epilepsy: Case reports based on ictal source localization analysis

Bratu

Nica²,

Oane³

et al. 2023

Front. Neurol.

Self Cite

View full text Add to dashboard Cite

Musicogenic epilepsy is a rare form of reflex epilepsy in which seizures are provoked by music. Different musicogenic stimuli have been identified: pleasant/unpleasant music or specific musical patterns. Several etiologies have been uncovered, such as focal cortical dysplasia, autoimmune encephalitis, tumors, or unspecific gliosis. In this article, we report two patients with musicogenic seizures. The first patient was diagnosed with structural temporal lobe epilepsy. Her seizures were elicited by music that she liked. Interictal and ictal video-electroencephalography (video-EEG) and signal analysis using independent component analysis revealed the right temporal lobe seizure onset extending over the neocortical regions. The patient underwent right temporal lobectomy (including the amygdala, the head, and the body of the hippocampus) and faced an Engel IA outcome 3 years post-surgery. The second patient was diagnosed with autoimmune temporal lobe epilepsy (GAD-65 antibodies). Her seizures were triggered by contemporary hit radio songs without any personal emotional significance. Interictal and ictal video-electroencephalography (video-EEG) and independent component analysis highlighted the left temporal lobe seizure onset extending over the neocortical regions. Intravenous immunoglobulin therapy was initiated, and the patient became seizure-free at 1 year. In conclusion, musicogenic seizures may be elicited by various auditory stimuli, the presence or absence of an emotional component offering an additional clue for the underlying network pathophysiology. Furthermore, in such cases, the use of independent component analysis of the scalp EEG signals proves useful in revealing the location of the seizure generator, and our findings point toward the temporal lobe, both mesial and neocortical regions.

show abstract

Extracting seizure onset from surface EEG with independent component analysis: Insights from simultaneous scalp and intracerebral EEG

Cited by 18 publications

References 42 publications

Neuromorphic Architecture Accelerated Automated Seizure Detection in Multi-Channel Scalp EEG

Neuromorphic Architecture Accelerated Automated Seizure Detection in Multi-Channel Scalp EEG

Validating EEG source imaging using intracranial electrical stimulation

Musicogenic seizures in temporal lobe epilepsy: Case reports based on ictal source localization analysis

Contact Info

Product

Resources

About