Deep neural networks constrained by neural mass models improve electrophysiological source imaging of spatiotemporal brain dynamics.

Many efforts have been made to image the spatiotemporal electrical activity of the brain with the purpose of mapping its function and dysfunction as well as aiding the management of brain disorders. Here, we propose a non-conventional deep learning-based source imaging framework (DeepSIF) that provides robust and precise spatiotemporal estimates of underlying brain dynamics from noninvasive high-density electroencephalography (EEG) recordings. DeepSIF employs synthetic training data generated by biophysical models capable of modeling mesoscale brain dynamics. The rich characteristics of underlying brain sources are embedded in the realistic training data and implicitly learned by DeepSIF networks, avoiding complications associated with explicitly formulating and tuning priors in an optimization problem, as often is the case in conventional source imaging approaches. The performance of DeepSIF is evaluated by 1) a series of numerical experiments, 2) imaging sensory and cognitive brain responses in a total of 20 healthy subjects from three public datasets, and 3) rigorously validating DeepSIF's capability in identifying epileptogenic regions in a cohort of 20 drug-resistant epilepsy patients by comparing DeepSIF results with invasive measurements and surgical resection outcomes. DeepSIF demonstrates robust and excellent performance, producing results that are concordant with common neuroscience knowledge about sensory and cognitive information processing as well as clinical findings about the location and extent of the epileptogenic tissue and outperforming conventional source imaging methods. The DeepSIF method, as a data-driven imaging framework, enables efficient and effective high-resolution functional imaging of spatiotemporal brain dynamics, suggesting its wide applicability and value to neuroscience research and clinical applications.

Source imaging method based on diagonal covariance bases and its applications to OPM-MEG.

Magnetoencephalography (MEG) is a noninvasive imaging technique used in neuroscience and clinical research. The source estimation of MEG involves solving a highly underdetermined inverse problem, which requires additional constraints to restrict the solution space. Traditional methods tend to obscure the extent of the sources. However, an accurate estimation of the source extent is important for studying brain activity or preoperatively estimating pathogenic regions. To improve the estimation accuracy of the extended source extent, the spatial constraint of sources is employed in the Bayesian framework. For example, the source is decomposed into a linear combination of validated spatial basis functions, which is proved to improve the source imaging accuracy. In this work, we further construct the spatial properties of the source using the diagonal covariance bases (DCB), which we summarize as the source imaging method SI-DCB. In this approach, specifically, the covariance matrix of the spatial coefficients is modeled as a weighted combination of diagonal covariance basis functions. The convex analysis is used to estimate noise and model parameters under the Bayesian framework. Extensive numerical simulations showed that SI-DCB outperformed five benchmark methods in accurately estimating the location and extent of patch sources. The effectiveness of SI-DCB was verified through somatosensory stimulation experiments performed on a 31-channel OPM-MEG system. The SI-DCB correctly identified the source area where each brain response occurred. The superior performance of SI-DCB suggests that it can provide a template approach for improving the accuracy of source extent estimations under a sparse Bayesian framework.

XDL-ESI: Electrophysiological Sources Imaging via explainable deep learning framework with validation on simultaneous EEG and iEEG.

Electroencephalography (EEG) or Magnetoencephalography (MEG) source imaging aims to estimate the underlying activated brain sources to explain the observed EEG/MEG recordings. Solving the inverse problem of EEG/MEG Source Imaging (ESI) is challenging due to its ill-posed nature. To achieve a unique solution, it is essential to apply sophisticated regularization constraints to restrict the solution space. Traditionally, the design of regularization terms is based on assumptions about the spatiotemporal structure of the underlying source dynamics. In this paper, we propose a novel paradigm for ESI via an Explainable Deep Learning framework, termed as XDL-ESI, which connects the iterative optimization algorithm with deep learning architecture by unfolding the iterative updates with neural network modules. The proposed framework has the advantages of (1) establishing a data-driven approach to model the source solution structure instead of using hand-crafted regularization terms; (2) improving the robustness of source solutions by introducing a topological loss that leverages the geometric spatial information applying varying penalties on distinct localization errors; (3) improving the reconstruction efficiency and interpretability as it inherits the advantages from both the iterative optimization algorithms (interpretability) and deep learning approaches (function approximation). The proposed XDL-ESI framework provides an efficient, accurate, and interpretable paradigm to solve the ESI inverse problem with satisfactory performance in both simulated data and real clinical data. Specially, this approach is further validated using simultaneous EEG and intracranial EEG (iEEG).

Electrophysiological brain imaging based on simulation-driven deep learning in the context of epilepsy.

Identifying the location, the spatial extent and the electrical activity of distributed brain sources in the context of epilepsy through ElectroEncephaloGraphy (EEG) recordings is a challenging task because of the highly ill-posed nature of the underlying Electrophysiological Source Imaging (ESI) problem. To guarantee a unique solution, most existing ESI methods pay more attention to solve this inverse problem by imposing physiological constraints. This paper proposes an efficient ESI approach based on simulation-driven deep learning. Epileptic High-resolution 256-channels scalp EEG (Hr-EEG) signals are simulated in a realistic manner to train the proposed patient-specific model. More particularly, a computational neural mass model developed in our team is used to generate the temporal dynamics of the activity of each dipole while the forward problem is solved using a patient-specific three-shell realistic head model and the boundary element method. A Temporal Convolutional Network (TCN) is considered in the proposed model to capture local spatial patterns. To enable the model to observe the EEG signals from different scale levels, the multi-scale strategy is leveraged to capture the overall features and fine-grain features by adjusting the convolutional kernel size. Then, the Long Short-Term Memory (LSTM) is used to extract temporal dependencies among the computed spatial features. The performance of the proposed method is evaluated through three different scenarios of realistic synthetic interictal Hr-EEG data as well as on real interictal Hr-EEG data acquired in three patients with drug-resistant partial epilepsy, during their presurgical evaluation. A performance comparison study is also conducted with two other deep learning-based methods and four classical ESI techniques. The proposed model achieved a Dipole Localization Error (DLE) of 1.39 and Normalized Hamming Distance (NHD) of 0.28 in the case of one patch with SNR of 10 dB. In the case of two uncorrelated patches with an SNR of 10 dB, obtained DLE and NHD were respectively 1.50 and 0.28. Even in the more challenging scenario of two correlated patches with an SNR of 10 dB, the proposed approach still achieved a DLE of 3.74 and an NHD of 0.43. The results obtained on simulated data demonstrate that the proposed method outperforms the existing methods for different signal-to-noise and source configurations. The good behavior of the proposed method is also confirmed on real interictal EEG data. The robustness with respect to noise makes it a promising and alternative tool to localize epileptic brain areas and to reconstruct their electrical activities from EEG signals.

Further characterisation of late somatosensory evoked potentials using electroencephalogram and magnetoencephalogram source imaging.

Beside the well-documented involvement of secondary somatosensory area, the cortical network underlying late somatosensory evoked potentials (P60/N60 and P100/N100) is still unknown. Electroencephalogram and magnetoencephalogram source imaging were performed to further investigate the origin of the brain cortical areas involved in late somatosensory evoked potentials, using sensory inputs of different strengths and by testing the correlation between cortical sources. Simultaneous high-density electroencephalograms and magnetoencephalograms were performed in 19 participants, and electrical stimulation was applied to the median nerve (wrist level) at intensity between 1.5 and 9 times the perceptual threshold. Source imaging was undertaken to map the stimulus-induced brain cortical activity according to each individual brain magnetic resonance imaging, during three windows of analysis covering early and late somatosensory evoked potentials. Results for P60/N60 and P100/N100 were compared with those for P20/N20 (early response). According to literature, maximal activity during P20/N20 was found in central sulcus contralateral to stimulation site. During P60/N60 and P100/N100, activity was observed in contralateral primary sensorimotor area, secondary somatosensory area (on both hemispheres) and premotor and multisensory associative cortices. Late responses exhibited similar characteristics but different from P20/N20, and no significant correlation was found between early and late generated activities. Specific clusters of cortical activities were activated with specific input/output relationships underlying early and late somatosensory evoked potentials. Cortical networks, partly common to and distinct from early somatosensory responses, contribute to late responses, all participating in the complex somatosensory brain processing.

EEG/MEG source imaging of deep brain activity within the maximum entropy on the mean framework: Simulations and validation in epilepsy.

Electro/Magneto-EncephaloGraphy (EEG/MEG) source imaging (EMSI) of epileptic activity from deep generators is often challenging due to the higher sensitivity of EEG/MEG to superficial regions and to the spatial configuration of subcortical structures. We previously demonstrated the ability of the coherent Maximum Entropy on the Mean (cMEM) method to accurately localize the superficial cortical generators and their spatial extent. Here, we propose a depth-weighted adaptation of cMEM to localize deep generators more accurately. These methods were evaluated using realistic MEG/high-density EEG (HD-EEG) simulations of epileptic activity and actual MEG/HD-EEG recordings from patients with focal epilepsy. We incorporated depth-weighting within the MEM framework to compensate for its preference for superficial generators. We also included a mesh of both hippocampi, as an additional deep structure in the source model. We generated 5400 realistic simulations of interictal epileptic discharges for MEG and HD-EEG involving a wide range of spatial extents and signal-to-noise ratio (SNR) levels, before investigating EMSI on clinical HD-EEG in 16 patients and MEG in 14 patients. Clinical interictal epileptic discharges were marked by visual inspection. We applied three EMSI methods: cMEM, depth-weighted cMEM and depth-weighted minimum norm estimate (MNE). The ground truth was defined as the true simulated generator or as a drawn region based on clinical information available for patients. For deep sources, depth-weighted cMEM improved the localization when compared to cMEM and depth-weighted MNE, whereas depth-weighted cMEM did not deteriorate localization accuracy for superficial regions. For patients' data, we observed improvement in localization for deep sources, especially for the patients with mesial temporal epilepsy, for which cMEM failed to reconstruct the initial generator in the hippocampus. Depth weighting was more crucial for MEG (gradiometers) than for HD-EEG. Similar findings were found when considering depth weighting for the wavelet extension of MEM. In conclusion, depth-weighted cMEM improved the localization of deep sources without or with minimal deterioration of the localization of the superficial sources. This was demonstrated using extensive simulations with MEG and HD-EEG and clinical MEG and HD-EEG for epilepsy patients.

Connectome spectrum electromagnetic tomography: A method to reconstruct electrical brain source networks at high-spatial resolution.

Connectome spectrum electromagnetic tomography (CSET) combines diffusion MRI-derived structural connectivity data with well-established graph signal processing tools to solve the M/EEG inverse problem. Using simulated EEG signals from fMRI responses, and two EEG datasets on visual-evoked potentials, we provide evidence supporting that (i) CSET captures realistic neurophysiological patterns with better accuracy than state-of-the-art methods, (ii) CSET can reconstruct brain responses more accurately and with more robustness to intrinsic noise in the EEG signal. These results demonstrate that CSET offers high spatio-temporal accuracy, enabling neuroscientists to extend their research beyond the current limitations of low sampling frequency in functional MRI and the poor spatial resolution of M/EEG.

High-density electroencephalographic functional networks in genetic generalized epilepsy: Preserved whole-brain topology hides local reorganization.

OBJECTIVE: Genetic generalized epilepsy (GGE) accounts for approximately 20% of adult epilepsy cases and is considered a disorder of large brain networks, involving both hemispheres. Most studies have not shown any difference in functional whole-brain network topology when compared to healthy controls. Our objective was to examine whether this preserved global network topology could hide local reorganizations that balance out at the global network level. METHODS: We recorded high-density electroencephalograms from 20 patients and 20 controls, and reconstructed the activity of 118 regions. We computed functional connectivity in windows free of interictal epileptiform discharges in broad, delta, theta, alpha, and beta frequency bands, characterized the network topology, and used the Hub Disruption Index (HDI) to quantify the topological reorganization. We examined the generalizability of our results by reproducing a 25-electrode clinical system. RESULTS: Our study did not reveal any significant change in whole-brain network topology among GGE patients. However, the HDI was significantly different between patients and controls in all frequency bands except alpha (p < .01, false discovery rate [FDR] corrected, d < -1), and accompanied by an increase in connectivity in the prefrontal regions and default mode network. This reorganization suggests that regions that are important in transferring the information in controls were less so in patients. Inversely, the crucial regions in patients are less so in controls. These findings were also found in delta and theta frequency bands when using 25 electrodes (p < .001, FDR corrected, d < -1). SIGNIFICANCE: In GGE patients, the overall network topology is similar to that of healthy controls but presents a balanced local topological reorganization. This reorganization causes the prefrontal areas and default mode network to be more integrated and segregated, which may explain executive impairment associated with GGE. Additionally, the reorganization distinguishes patients from controls even when using 25 electrodes, suggesting its potential use as a diagnostic tool.

Electroencephalographic source imaging of spikes with concurrent high-frequency oscillations is concordant with the clinical ground truth.

OBJECTIVE: Epilepsy raises critical challenges to accurately localize the epileptogenic zone (EZ) to guide presurgical planning. Previous research has suggested that interictal spikes overlapping with high-frequency oscillations, referred to here as pSpikes, serve as a reliable biomarker for EZ estimation, but there remains a question as to whether and to how pSpikes perform as compared to other types of epileptic spikes. This study aims to address this question by investigating the source imaging capabilities of pSpikes alongside other spike types. METHODS: A total of 2819 interictal spikes from 76-channel scalp electroencephalography (EEG) were analyzed in a cohort of 24 drug-resistant focal epilepsy patients. All patients received surgical resection, and 16 were declared seizure-free based on at least 1 year of postoperative follow-up. A recently developed electrophysiological source imaging algorithm-fast spatiotemporal iteratively reweighted edge sparsity (FAST-IRES)-was used for source imaging of the detected interictal spikes. The performance of 217 pSpikes was compared with 772 nSpikes (spikes with irregular high-frequency activations), 1830 rSpikes (spikes with no high-frequency activity), and all 2819 aSpikes (all interictal spikes). RESULTS: The localization and extent estimation using pSpikes are concordant with the clinical ground truth; using pSpikes yields the best performance compared with nSpikes, rSpikes, and conventional spike imaging (aSpikes). For multiple spike type seizure-free patients, the mean localization error for pSpike imaging was 6.8 mm, compared with 15.0 mm for aSpikes. The sensitivity, precision, and specificity were .41, .67, and .93 for pSpikes compared with .32, .48, and .93 for aSpikes. SIGNIFICANCE: These results demonstrate the merits of noninvasive EEG source localization, and that (1) pSpike is a superior biomarker, outperforming conventional spike imaging for the localization of epileptic sources, and especially those with multiple irritative zones; and (2) FAST-IRES provides accurate source estimation that is highly concordant with clinical ground truth, even in situations of single spike analysis with low signal-to-noise ratio.

Interictal EEG source connectivity to localize the epileptogenic zone in patients with drug-resistant epilepsy: A machine learning approach.

OBJECTIVE: To deconstruct the epileptogenic networks of patients with drug-resistant epilepsy (DRE) using source functional connectivity (FC) analysis; unveil the FC biomarkers of the epileptogenic zone (EZ); and develop machine learning (ML) models to estimate the EZ using brief interictal electroencephalography (EEG) data. METHODS: We analyzed scalp EEG from 50 patients with DRE who had surgery. We reconstructed the activity (electrical source imaging [ESI]) of virtual sensors (VSs) across the whole cortex and computed FC separately for epileptiform and non-epileptiform EEG epochs (with or without spikes). In patients with good outcome (Engel 1a), four cortical regions were defined: EZ (resection) and three non-epileptogenic zones (NEZs) in the same and opposite hemispheres. Region-specific FC features in six frequency bands and three spatial ranges (long, short, inner) were compared between regions (Wilcoxon sign-rank). We developed ML classifiers to identify the VSs in the EZ using VS-specific FC features. Cross-validation was performed using good outcome data. Performance was compared with poor outcomes and interictal spike localization. RESULTS: FC differed between EZ and NEZs (p < .05) during non-epileptiform and epileptiform epochs, showing higher FC in the EZ than its homotopic contralateral NEZ. During epileptiform epochs, the NEZ in the epileptogenic hemisphere showed higher FC than its contralateral NEZ. In good outcome patients, the ML classifiers reached 75% accuracy to the resection (91% sensitivity; 74% specificity; distance from EZ: 38 mm) using epileptiform epochs (gamma and beta frequency bands) and 62% accuracy using broadband non-epileptiform epochs, both outperforming spike localization (accuracy = 47%; p < .05; distance from EZ: 57 mm). Lower performance was seen in poor outcomes. SIGNIFICANCE: We present an FC approach to extract EZ biomarkers from brief EEG data. Increased FC in various frequencies characterized the EZ during epileptiform and non-epileptiform epochs. FC-based ML models identified the resection better in good than poor outcome patients, demonstrating their potential for presurgical use in pediatric DRE.

Semiautomated electric source imaging determines epileptogenicity of encephaloceles in temporal lobe epilepsy.

OBJECTIVE: We aimed to assess the ability of semiautomated electric source imaging (ESI) from long-term video-electroencephalographic (EEG) monitoring (LTM) to determine the epileptogenicity of temporopolar encephaloceles (TEs) in patients with temporal lobe epilepsy. METHODS: We conducted a retrospective study involving 32 temporal lobe epilepsy patients with TEs as potentially epileptogenic lesions in structural magnetic resonance imaging scans. Findings were validated through invasive intracerebral stereo-EEG in six of 32 patients and postsurgical outcome after tailored resection of the TE in 17 of 32 patients. LTM (mean duration = 6 days) was performed using the 10/20 system with additional T1/T2 for all patients and sphenoidal electrodes in 23 of 32 patients. Semiautomated detection and clustering of interictal epileptiform discharges (IEDs) were carried out to create IED types. ESI was performed on the averages of the two most frequent IED types per patient, utilizing individual head models, and two independent inverse methods (sLORETA [standardized low-resolution brain electromagnetic tomography], MUSIC [multiple signal classification]). ESI maxima concordance and propagation in spatial relation to TEs were quantified for sources with good signal quality (signal-to-noise ratio > 2, explained signal > 60%). RESULTS: ESI maxima correctly colocalized with a TE in 20 of 32 patients (62.5%) either at the onset or half-rising flank of at least one IED type per patient. ESI maxima showed propagation from the temporal pole to other temporal or extratemporal regions in 14 of 32 patients (44%), confirming propagation originating in the area of the TE. The findings from both inverse methods validated each other in 14 of 20 patients (70%), and sphenoidal electrodes exhibited the highest signal amplitudes in 17 of 23 patients (74%). The concordance of ESI with the TE predicted a seizure-free postsurgical outcome (Engel I vs. >I) with a diagnostic odds ratio of 2.1. SIGNIFICANCE: Semiautomated ESI from LTM often successfully identifies the epileptogenicity of TEs and the IED onset zone within the area of the TEs. Additionally, it shows potential predictive power for postsurgical outcomes in these patients.

The impact of CSF-filled cavities on scalp EEG and its implications.

Previous studies have found electroencephalogram (EEG) amplitude and scalp topography differences between neurotypical and neurological/neurosurgical groups, being interpreted at the cognitive level. However, these comparisons are invariably accompanied by anatomical changes. Critical to EEG are the so-called volume currents, which are affected by the spatial distribution of the different tissues in the head. We investigated the effect of cerebrospinal fluid (CSF)-filled cavities on simulated EEG scalp data. We simulated EEG scalp potentials for known sources using different volume conduction models: a reference model (i.e., unlesioned brain) and models with realistic CSF-filled cavities gradually increasing in size. We used this approach for a single source close or far from the CSF-lesion cavity, and for a scenario with a distributed configuration of sources (i.e., a "cognitive event-related potential effect"). The magnitude and topography errors between the reference and lesion models were quantified. For the single-source simulation close to the lesion, the CSF-filled lesion modulated signal amplitude with more than 17% magnitude error and topography with more than 9% topographical error. Negligible modulation was found for the single source far from the lesion. For the multisource simulations of the cognitive effect, the CSF-filled lesion modulated signal amplitude with more than 6% magnitude error and topography with more than 16% topography error in a nonmonotonic fashion. In conclusion, the impact of a CSF-filled cavity cannot be neglected for scalp-level EEG data. Especially when group-level comparisons are made, any scalp-level attenuated, aberrant, or absent effects are difficult to interpret without considering the confounding effect of CSF.

Spike propagation mapping reveals effective connectivity and predicts surgical outcome in epilepsy.

Neurosurgical intervention is the best available treatment for selected patients with drug resistant epilepsy. For these patients, surgical planning requires biomarkers that delineate the epileptogenic zone, the brain area that is indispensable for the generation of seizures. Interictal spikes recorded with electrophysiological techniques are considered key biomarkers of epilepsy. Yet, they lack specificity, mostly because they propagate across brain areas forming networks. Understanding the relationship between interictal spike propagation and functional connections among the involved brain areas may help develop novel biomarkers that can delineate the epileptogenic zone with high precision. Here, we reveal the relationship between spike propagation and effective connectivity among onset and areas of spread and assess the prognostic value of resecting these areas. We analysed intracranial EEG data from 43 children with drug resistant epilepsy who underwent invasive monitoring for neurosurgical planning. Using electric source imaging, we mapped spike propagation in the source domain and identified three zones: onset, early-spread and late-spread. For each zone, we calculated the overlap and distance from surgical resection. We then estimated a virtual sensor for each zone and the direction of information flow among them via Granger causality. Finally, we compared the prognostic value of resecting these zones, the clinically-defined seizure onset zone and the spike onset on intracranial EEG channels by estimating their overlap with resection. We observed a spike propagation in source space for 37 patients with a median duration of 95 ms (interquartile range: 34-206), a spatial displacement of 14 cm (7.5-22 cm) and a velocity of 0.5 m/s (0.3-0.8 m/s). In patients with good surgical outcome (25 patients, Engel I), the onset had higher overlap with resection [96% (40-100%)] than early-spread [86% (34-100%), P = 0.01] and late-spread [59% (12-100%), P = 0.002], and it was also closer to resection than late-spread [5 mm versus 9 mm, P = 0.007]. We found an information flow from onset to early-spread in 66% of patients with good outcomes, and from early-spread to onset in 50% of patients with poor outcome. Finally, resection of spike onset, but not area of spike spread or the seizure onset zone, predicted outcome with positive predictive value of 79% and negative predictive value of 56% (P = 0.04). Spatiotemporal mapping of spike propagation reveals information flow from onset to areas of spread in epilepsy brain. Surgical resection of the spike onset disrupts the epileptogenic network and may render patients with drug resistant epilepsy seizure-free without having to wait for a seizure to occur during intracranial monitoring.

EEG Spatial-temporal Dynamics of Resting-state Activity in Young Women with Anorexia Nervosa: Preliminary Evidence.

The aim of this study was to provide preliminary evidence on temporal dynamics of resting-state brain networks in youth with anorexia nervosa (AN) using electroencephalography (EEG). Resting-state EEG data were collected in 18 young women with AN and 18 healthy controls (HC). Between-group differences in brain networks were assessed using microstates analyses. Five microstates were identified across all subjects (A, B, C, D, E). Using a single set of maps representative of the whole dataset, group differences were identified for microstates A, C, and E. A common-for-all template revealed a relatively high degree of consistency in results for reduced time coverage of microstate C, but also an increased presence of microstate class E. AN and HC had different microstate transition probabilities, largely involving microstate A. Using LORETA, for microstate D, we found that those with AN had augmented activations in the left frontal inferior operculum, left insula, and bilateral paracentral lobule, compared with HC. For microstate E, AN had augmented activations in the para-hippocampal gyrus, caudate, pallidum, cerebellum, and cerebellar vermis. Our findings suggest altered microstates in young women with AN associated with integration of sensory and bodily signals, monitoring of internal/external mental states, and self-referential processes. Future research should examine how EEG-derived microstates could be applied to develop diagnostic and prognostic models of AN.

Sleep Spindle Generation Before and After Epilepsy Surgery: A Source Imaging Study in Children with Drug-Resistant Epilepsy.

INTRODUCTION: Literature lacks studies investigating the cortical generation of sleep spindles in drug-resistant epilepsy (DRE) and how they evolve after resection of the epileptogenic zone (EZ). Here, we examined sleep EEGs of children with focal DRE who became seizure-free after focal epilepsy surgery, and aimed to investigate the changes in the spindle generation before and after the surgery using low-density scalp EEG and electrical source imaging (ESI). METHODS: We analyzed N2-sleep EEGs from 19 children with DRE before and after surgery. We identified slow (8-12 Hz) and fast spindles (13-16 Hz), computed their spectral features and cortical generators through ESI and computed their distance from the EZ and irritative zone (IZ). We performed two-way ANOVA testing the effect of spindle type (slow vs. fast) and surgical phase (pre-surgery vs. post-surgery) on each feature. RESULTS: Power, frequency and cortical activation of slow spindles increased after surgery (p < 0.005), while this was not seen for fast spindles. Before surgery, the cortical generators of slow spindles were closer to the EZ (57.3 vs. 66.2 mm, p = 0.007) and IZ (41.3 vs. 55.5 mm, p = 0.02) than fast spindle generators. CONCLUSIONS: Our data indicate alterations in the EEG slow spindles after resective epilepsy surgery. Fast spindle generation on the contrary did not change after surgery. Although the study is limited by its retrospective nature, lack of healthy controls, and reduced cortical spatial sampling, our findings suggest a spatial relationship between the slow spindles and the epileptogenic generators.

Noninvasive high-frequency oscillations riding spikes delineates epileptogenic sources.

High-frequency oscillations (HFOs) are a promising biomarker for localizing epileptogenic brain and guiding successful neurosurgery. However, the utility and translation of noninvasive HFOs, although highly desirable, is impeded by the difficulty in differentiating pathological HFOs from nonepileptiform high-frequency activities and localizing the epileptic tissue using noninvasive scalp recordings, which are typically contaminated with high noise levels. Here, we show that the consistent concurrence of HFOs with epileptiform spikes (pHFOs) provides a tractable means to identify pathological HFOs automatically, and this in turn demarks an epileptiform spike subgroup with higher epileptic relevance than the other spikes in a cohort of 25 temporal epilepsy patients (including a total of 2,967 interictal spikes and 1,477 HFO events). We found significant morphological distinctions of HFOs and spikes in the presence/absence of this concurrent status. We also demonstrated that the proposed pHFO source imaging enhanced localization of epileptogenic tissue by 162% (∼5.36 mm) for concordance with surgical resection and by 186% (∼12.48 mm) with seizure-onset zone determined by invasive studies, compared to conventional spike imaging, and demonstrated superior congruence with the surgical outcomes. Strikingly, the performance of spike imaging was selectively boosted by the presence of spikes with pHFOs, especially in patients with multitype spikes. Our findings suggest that concurrent HFOs and spikes reciprocally discriminate pathological activities, providing a translational tool for noninvasive presurgical diagnosis and postsurgical evaluation in vulnerable patients.

µ-STAR: A novel framework for spatio-temporal M/EEG source imaging optimized by microstates.

Source imaging of Electroencephalography (EEG) and Magnetoencephalography (MEG) provides a noninvasive way of monitoring brain activities with high spatial and temporal resolution. In order to address this highly ill-posed problem, conventional source imaging models adopted spatio-temporal constraints that assume spatial stability of the source activities, neglecting the transient characteristics of M/EEG. In this work, a novel source imaging method µ-STAR that includes a microstate analysis and a spatio-temporal Bayesian model was introduced to address this problem. Specifically, the microstate analysis was applied to achieve automatic determination of time window length with quasi-stable source activity pattern for optimal reconstruction of source dynamics. Then a user-specific spatial prior and data-driven temporal basis functions were utilized to characterize the spatio-temporal information of sources within each state. The solution of the source reconstruction was obtained through a computationally efficient algorithm based upon variational Bayesian and convex analysis. The performance of the µ-STAR was first assessed through numerical simulations, where we found that the determination and inclusion of optimal temporal length in the spatio-temporal prior significantly improved the performance of source reconstruction. More importantly, the µ-STAR model achieved robust performance under various settings (i.e., source numbers/areas, SNR levels, and source depth) with fast convergence speed compared with five widely-used benchmark models (including wMNE, STV, SBL, BESTIES, & SI-STBF). Additional validations on real data were then performed on two publicly-available datasets (including block-design face-processing ERP and continuous resting-state EEG). The reconstructed source activities exhibited spatial and temporal neurophysiologically plausible results consistent with previously-revealed neural substrates, thereby further proving the feasibility of the µ-STAR model for source imaging in various applications.

Validating MEG source imaging of resting state oscillatory patterns with an intracranial EEG atlas.

BACKGROUND: Magnetoencephalography (MEG) is a widely used non-invasive tool to estimate brain activity with high temporal resolution. However, due to the ill-posed nature of the MEG source imaging (MSI) problem, the ability of MSI to identify accurately underlying brain sources along the cortical surface is still uncertain and requires validation. METHOD: We validated the ability of MSI to estimate the background resting state activity of 45 healthy participants by comparing it to the intracranial EEG (iEEG) atlas (https://mni-open-ieegatlas. RESEARCH: mcgill.ca/). First, we applied wavelet-based Maximum Entropy on the Mean (wMEM) as an MSI technique. Next, we converted MEG source maps into intracranial space by applying a forward model to the MEG-reconstructed source maps, and estimated virtual iEEG (ViEEG) potentials on each iEEG channel location; we finally quantitatively compared those with actual iEEG signals from the atlas for 38 regions of interest in the canonical frequency bands. RESULTS: The MEG spectra were more accurately estimated in the lateral regions compared to the medial regions. The regions with higher amplitude in the ViEEG than in the iEEG were more accurately recovered. In the deep regions, MEG-estimated amplitudes were largely underestimated and the spectra were poorly recovered. Overall, our wMEM results were similar to those obtained with minimum norm or beamformer source localization. Moreover, the MEG largely overestimated oscillatory peaks in the alpha band, especially in the anterior and deep regions. This is possibly due to higher phase synchronization of alpha oscillations over extended regions, exceeding the spatial sensitivity of iEEG but detected by MEG. Importantly, we found that MEG-estimated spectra were more comparable to spectra from the iEEG atlas after the aperiodic components were removed. CONCLUSION: This study identifies brain regions and frequencies for which MEG source analysis is likely to be reliable, a promising step towards resolving the uncertainty in recovering intracerebral activity from non-invasive MEG studies.

Sensor array design of optically pumped magnetometers for accurately estimating source currents.

An optically pumped magnetometer (OPM) is a new generation of magnetoencephalography (MEG) devices that is small, light, and works at room temperature. Due to these characteristics, OPMs enable flexible and wearable MEG systems. On the other hand, if we have a limited number of OPM sensors, we need to carefully design their sensor arrays depending on our purposes and regions of interests (ROIs). In this study, we propose a method that designs OPM sensor arrays for accurately estimating the cortical currents at the ROIs. Based on the resolution matrix of minimum norm estimate (MNE), our method sequentially determines the position of each sensor to optimize its inverse filter pointing to the ROIs and suppressing the signal leakage from the other areas. We call this method the Sensor array Optimization based on Resolution Matrix (SORM). We conducted simple and realistic simulation tests to evaluate its characteristics and efficacy for real OPM-MEG data. SORM designed the sensor arrays so that their leadfield matrices had high effective ranks as well as high sensitivities to ROIs. Although SORM is based on MNE, the sensor arrays designed by SORM were effective not only when we estimated the cortical currents by MNE but also when we did so by other methods. With real OPM-MEG data we confirmed its validity for real data. These analyses suggest that SORM is especially useful when we want to accurately estimate ROIs' activities with a limited number of OPM sensors, such as brain-machine interfaces and diagnosing brain diseases.

Deep learning based source imaging provides strong sublobar localization of epileptogenic zone from MEG interictal spikes.

Electromagnetic source imaging (ESI) offers unique capability of imaging brain dynamics for studying brain functions and aiding the clinical management of brain disorders. Challenges exist in ESI due to the ill-posedness of the inverse problem and thus the need of modeling the underlying brain dynamics for regularizations. Advances in generative models provide opportunities for more accurate and realistic source modeling that could offer an alternative approach to ESI for modeling the underlying brain dynamics beyond equivalent physical source models. However, it is not straightforward to explicitly formulate the knowledge arising from these generative models within the conventional ESI framework. Here we investigate a novel source imaging framework based on mesoscale neuronal modeling and deep learning (DL) that can learn the sensor-source mapping relationship directly from MEG data for ESI. Two DL-based ESI models were trained based on data generated by neural mass models and either generic or personalized head models. The robustness of the two DL models was evaluated by systematic computer simulations and clinical validation in a cohort of 29 drug-resistant focal epilepsy patients who underwent intracranial EEG (iEEG) evaluation or surgical resection. Results estimated from pre-operative MEG interictal spikes were quantified using the overlap with resection regions and the distance to the seizure-onset zone (SOZ) defined by iEEG recordings. The DL-based ESI provided robust results when no personalized head geometry is considered, reaching a spatial dispersion of 21.90 ± 19.03 mm, sublobar concordance of 83 %, and sublobar sensitivity and specificity of 66 and 97 % respectively. When using personalized head geometry derived from individual patients' MRI in the training data, personalized DL-based ESI model can further improve the performance and reached a spatial dispersion of 8.19 ± 8.14 mm, sublobar concordance of 93 %, and sublobar sensitivity and specificity of 77 and 99 % respectively. When compared to the SOZ, the localization error of the personalized approach is 15.78 ± 5.54 mm, outperforming the conventional benchmarks. This work demonstrates that combining generative models and deep learning enables an accurate and robust imaging of epileptogenic zone from MEG recordings with strong sublobar precision, suggesting its added value to enhancing MEG source localization and imaging, and to epilepsy source localization and other clinical applications.