The SESAMEEG package: a probabilistic tool for source localization and uncertainty quantification in M/EEG.

Source localization from M/EEG data is a fundamental step in many analysis pipelines, including those aiming at clinical applications such as the pre-surgical evaluation in epilepsy. Among the many available source localization algorithms, SESAME (SEquential SemiAnalytic Montecarlo Estimator) is a Bayesian method that distinguishes itself for several good reasons: it is highly accurate in localizing focal sources with comparably little sensitivity to input parameters; it allows the quantification of the uncertainty of the reconstructed source(s); it accepts user-defined a priori high- and low-probability search regions in input; it can localize the generators of neural oscillations in the frequency domain. Both a Python and a MATLAB implementation of SESAME are available as open-source packages under the name of SESAMEEG and are well integrated with the main software packages used by the M/EEG community; moreover, the algorithm is part of the commercial software BESA Research (from version 7.0 onwards). While SESAMEEG is arguably simpler to use than other source modeling methods, it has a much richer output that deserves to be described thoroughly. In this article, after a gentle mathematical introduction to the algorithm, we provide a complete description of the available output and show several use cases on experimental M/EEG data.

Ongoing EEG artifact correction using blind source separation.

OBJECTIVE: Analysis of the electroencephalogram (EEG) for epileptic spike and seizure detection or brain-computer interfaces can be severely hampered by the presence of artifacts. The aim of this study is to describe and evaluate a fast automatic algorithm for ongoing correction of artifacts in continuous EEG recordings, which can be applied offline and online. METHODS: The automatic algorithm for ongoing correction of artifacts is based on fast blind source separation. It uses a sliding window technique with overlapping epochs and features in the spatial, temporal and frequency domain to detect and correct ocular, cardiac, muscle and powerline artifacts. RESULTS: The approach was validated in an independent evaluation study on publicly available continuous EEG data with 2035 marked artifacts. Validation confirmed that 88% of the artifacts could be removed successfully (ocular: 81%, cardiac: 84%, muscle: 98%, powerline: 100%). It outperformed state-of-the-art algorithms both in terms of artifact reduction rates and computation time. CONCLUSIONS: Fast ongoing artifact correction successfully removed a good proportion of artifacts, while preserving most of the EEG signals. SIGNIFICANCE: The presented algorithm may be useful for ongoing correction of artifacts, e.g., in online systems for epileptic spike and seizure detection or brain-computer interfaces.

Sensitivity of a 29-Channel MEG Source Montage.

In this paper, we study the performance of a source montage corresponding to 29 brain regions reconstructed from whole-head magnetoencephalographic (MEG) recordings, with the aim of facilitating the review of MEG data containing epileptiform discharges. Test data were obtained by superposing simulated signals from 100-nAm dipolar sources to a resting state MEG recording from a healthy subject. Simulated sources were placed systematically to different cortical locations for defining the optimal regularization for the source montage reconstruction and for assessing the detectability of the source activity from the 29-channel MEG source montage. The signal-to-noise ratio (SNR), computed for each source from the sensor-level and source-montage signals, was used as the evaluation parameter. Without regularization, the SNR from the simulated sources was larger in the sensor-level signals than in the source montage reconstructions. Setting the regularization to 2% increased the source montage SNR to the same level as the sensor-level SNR, improving the detectability of the simulated events from the source montage reconstruction. Sources producing a SNR of at least 15 dB were visually detectable from the source-montage signals. Such sources are located closer than about 75 mm from the MEG sensors, in practice covering all areas in the grey matter. The 29-channel source montage creates more focal signals compared to the sensor space and can significantly shorten the detection time of epileptiform MEG discharges for focus localization.

EEG-fMRI: Ballistocardiogram Artifact Reduction by Surrogate Method for Improved Source Localization.

For the analysis of simultaneous EEG-fMRI recordings, it is vital to use effective artifact removal tools. This applies in particular to the ballistocardiogram (BCG) artifact which is difficult to remove without distorting signals of interest related to brain activity. Here, we documented the use of surrogate source models to separate the artifact-related signals from brain signals with minimal distortion of the brain activity of interest. The artifact topographies used for surrogate separation were created automatically using principal components analysis (PCA-S) or by manual selection of artifact components utilizing independent components analysis (ICA-S). Using real resting-state data from 55 subjects superimposed with simulated auditory evoked potentials (AEP), both approaches were compared with three established BCG artifact removal methods: Blind Source Separation (BSS), Optimal Basis Set (OBS), and a mixture of both (OBS-ICA). Each method was evaluated for its applicability for ERP and source analysis using the following criteria: the number of events surviving artifact threshold scans, signal-to-noise ratio (SNR), error of source localization, and signal variance explained by the dipolar model. Using these criteria, PCA-S and ICA-S fared best overall, with highly significant differences to the established methods, especially in source localization. The PCA-S approach was also applied to a single subject Berger experiment performed in the MRI scanner. Overall, the removal of BCG artifacts by the surrogate methods provides a substantial improvement for the analysis of simultaneous EEG-fMRI data compared to the established methods.

Relative Source Power: A novel method for localizing epileptiform EEG discharges.

OBJECTIVE: To validate relative source power (RSP) imaging of extratemporal interictal epileptiform discharges (IEDs). METHODS: The accuracy of RSP was validated in a cohort of patients with extratemporal focal epilepsy and a confined epileptogenic lesion (<19 cm3) using distance to the lesion, concordance with resected area and postoperative outcome. Performance was compared with three conventional methods: voltage maps, equivalent current dipole and a distributed source model. RESULTS: Thirty-three of 41 consecutive patients (80%) had IED averages suitable for analysis. While the peak negativity in voltage maps localized above the epileptogenic lesion only in 18 cases, RSP-maps matched in 29 cases (88%, p < 0.0026). Source localization showed a median distance of 9.8 mm from the lesion. Source-regions with 20 mm radius included 98% of all source-to-lesion distances. In the 21 surgical cases, outcome showed a sensitivity of 82.35% and specificity of 50% without significant differences between the three source imaging methods. CONCLUSIONS: RSP-maps provide a rapid, intuitive and more accurate source estimation than voltage maps. At sublobar level, RSP localizes with an accuracy similar to conventional methods and results of previous studies. SIGNIFICANCE: The definition of a source region with 20 mm radius helps in guiding further exploration in extratemporal focal epilepsy.

LORETA With Cortical Constraint: Choosing an Adequate Surface Laplacian Operator.

Low resolution electromagnetic tomography (LORETA) is a well-known method for the solution of the l2-based minimization problem for EEG/MEG source reconstruction. LORETA with a volume-based source space is widely used and much effort has been invested in the theory and the application of the method in an experimental context. However, it is especially interesting to use anatomical prior knowledge and constrain the LORETA's solution to the cortical surface. This strongly reduces the number of unknowns in the inverse approach. Unlike the Laplace operator in the volume case with a rectangular and regular grid, the mesh is triangulated and highly irregular in the surface case. Thus, it is not trivial to choose or construct a Laplace operator (termed Laplace-Beltrami operator when applied to surfaces) that has the desired properties and takes into account the geometry of the mesh. In this paper, the basic methodology behind cortical LORETA is discussed and the method is applied for source reconstruction of simulated data using different Laplace-Beltrami operators in the smoothing term. The results achieved with the different operators are compared with respect to their accuracy using various measures. Conclusions about the choice of an appropriate operator are deduced from the results.

The Discontinuous Galerkin Finite Element Method for Solving the MEG and the Combined MEG/EEG Forward Problem.

In Electro- (EEG) and Magnetoencephalography (MEG), one important requirement of source reconstruction is the forward model. The continuous Galerkin finite element method (CG-FEM) has become one of the dominant approaches for solving the forward problem over the last decades. Recently, a discontinuous Galerkin FEM (DG-FEM) EEG forward approach has been proposed as an alternative to CG-FEM (Engwer et al., 2017). It was shown that DG-FEM preserves the property of conservation of charge and that it can, in certain situations such as the so-called skull leakages, be superior to the standard CG-FEM approach. In this paper, we developed, implemented, and evaluated two DG-FEM approaches for the MEG forward problem, namely a conservative and a non-conservative one. The subtraction approach was used as source model. The validation and evaluation work was done in statistical investigations in multi-layer homogeneous sphere models, where an analytic solution exists, and in a six-compartment realistically shaped head volume conductor model. In agreement with the theory, the conservative DG-FEM approach was found to be superior to the non-conservative DG-FEM implementation. This approach also showed convergence with increasing resolution of the hexahedral meshes. While in the EEG case, in presence of skull leakages, DG-FEM outperformed CG-FEM, in MEG, DG-FEM achieved similar numerical errors as the CG-FEM approach, i.e., skull leakages do not play a role for the MEG modality. In particular, for the finest mesh resolution of 1 mm sources with a distance of 1.59 mm from the brain-CSF surface, DG-FEM yielded mean topographical errors (relative difference measure, RDM%) of 1.5% and mean magnitude errors (MAG%) of 0.1% for the magnetic field. However, if the goal is a combined source analysis of EEG and MEG data, then it is highly desirable to employ the same forward model for both EEG and MEG data. Based on these results, we conclude that the newly presented conservative DG-FEM can at least complement and in some scenarios even outperform the established CG-FEM approaches in EEG or combined MEG/EEG source analysis scenarios, which motivates a further evaluation of DG-FEM for applications in bioelectromagnetism.

DC-EEG recordings of mindfulness.

OBJECTIVE: To assess the frequency spectrum of the normal waking human eyes-closed EEG while concentrating on a mental task. METHODS: Ten adult normal volunteers listened to a CD encouraging mindfulness for one hour and five minutes while their EEG was recorded on a 128 channel DC based ANT system. The software package BESA Research version 6.1 was used for data analysis. The data were subjected to topographic display, frequency as well as independent component analysis. RESULTS: Near-DC activity that extended beyond one hour, as well as rhythmic wave durations ranging from about 10 to 35min, was observed in all subjects. For this task the major topographic distribution was mainly in frontal near midline areas and the inferior portions of the hemispheres. CONCLUSIONS: The study demonstrated that rhythms below the infraslow band, as well as a near-DC component, exist in the normal human EEG. Their significance for health and disease now needs to be explored. SIGNIFICANCE: Since DC-based EEG/MEG systems are already in use by some laboratories, investigators are encouraged to include the exploration of these ultra-slow waves in the review of their data.

Magnetoencephalographic Infraslow Activity: A Feasibility Study.

PURPOSE: To explore if background infraslow activity (ISA) can be retrieved from archived magnetoencephalographic (MEG) recordings and its potential clinical relevance. METHODS: Archived recordings of 15 patients with epilepsy and 10 normal subjects were evaluated for MEG/EEG delta (0.5-3 Hz) and ISA (0.01-0.1 Hz). The data were obtained on a Neuromag/Elekta system with 204 planar gradiometers and 102 magnetometer sensors and also 60 EEG channels. To remove artifacts, all MEG files were temporal signal space separation filtered. The data were then analyzed with the BESA Research software. RESULTS: Infraslow activity was present in all files for MEG and EEG. Good concordance between EEG and MEG ISA was seen with delta for laterality and with clinical features. Delta frequencies were always less than 2 Hz. During sleep, an inverse relationship between delta and ISA occurred. With increasing depth of sleep, delta activity increased while ISA decreased and vice versa. Intermittent higher amplitude transients, arising from background, were also seen but their nature is at present unknown. Clinically relevant ictal onset baseline shifts were likewise observed. CONCLUSION: Infraslow activity is a normal segment of the cerebral electromagnetic frequency spectrum. It follows physiologic rules and can be related to areas of pathology. This is in accord with previously published EEG observations and further studies of this segment of the electromagnetic frequency spectrum for its origin and changes in health and disease are indicated.

Advanced tools for digital EEG review: virtual source montages, whole-head mapping, correlation, and phase analysis.

Digital EEG allows one to combine recorded EEG channels into new montages without the need to record new data. Using spherical splines, voltages can be estimated at any point on the head. This allows one to generate various montages with the recorded or virtual electrodes at standardized locations, to interpolate bad electrodes, and to generate topographic maps over the whole head. Simulations of EEG activity originating in various brain regions are used to illustrate the effects of known generators on various montages and on whole-head maps. Some properties of spatial filters are introduced, and it is shown how they can be used to develop source montages with signals that estimate the activity in specific brain regions. The usefulness and validity of a source montage designed to focus on temporal lobe activity is illustrated with simulations and examples of temporal lobe spikes and seizures. Additional tools such as cross-correlation among channels, fast Fourier transform, and phase maps are described. These tools are useful in estimating time lags between source channels and in interpreting propagating spike and seizure activity. In combination, these tools help to analyze and to enhance activities that may be hard to detect from the background scalp EEG in traditional montages.

How do children prepare to react? Imaging maturation of motor preparation and stimulus anticipation by late contingent negative variation.

Both the motor system and the frontal executive control system show a late maturation in humans which continues into school-age and even adolescence. We investigated the maturation of preparation processes towards a fast motor reaction in 74 healthy right-handed children aged 6 to 18 years and analyzed the topography of the late component of contingent negative variation (lCNV) in a 64-electrode high density sensor array. While adolescents from about 12 years on showed a bilaterally distributed centro-parietal maximum like adults do, younger children almost completely missed the negativity over the left central area contralaterally to the side of the anticipated movement. The reason, as revealed by current source density, was that only adolescents showed significant evoked activity of the left pre-/primary motor and supplementary/cingulate motor areas, while in contrast both age groups displayed significant current sinks over the right (ipsilateral) centro-temporal area and right posterior parietal cortex. Spatio-temporal source analysis confirmed that negativity over the right posterior parietal area could not be explained by a projection via volume conduction from frontal areas involved in motor preparation but represented an independent component with a different maturational course most likely related to sensory attention. Significant event-related desynchronization of alpha-power over the contralateral sensorimotor cortex was found in the younger age group, indicating that also 6- to 11-year-old children were engaged in motor preparation. Thus, the missing current sink over the contalateral sensorimotor cortex during late CNV in 6- to 11-year-old children might reflect the immaturity of a specific subcomponent of the motor preparation system which is related to evoked (late CNV) but not induced activity (alpha-ERD).

BESA source coherence: a new method to study cortical oscillatory coupling.

This paper introduces source coherence, a new method for the analysis of cortical coherence using noninvasive EEG and MEG data. Brain electrical source analysis (BESA) is applied to create a discrete multiple source model. This model is used as a source montage to transform the recorded data from sensor level into brain source space. This provides source waveforms of the modeled brain regions as a direct measure for their activities on a single trial basis. The source waveforms are transformed into time-frequency space using complex demodulation. Magnitude-squared coherence between the brain sources reveals oscillatory coupling between sources. This procedure allows one to separate the time-frequency content of different brain regions even if their activities severely overlap at the surface. Thus, source coherence overcomes problems of localization and interpretation that are inherent to coherence analysis at sensor level. The principle of source coherence is illustrated using an EEG recording of an error-related negativity as an example. In this experiment the subject performed a visuo-motor task. Source coherence analysis revealed dynamical linking between posterior and central areas within the gamma-band around the time of button press at a post-stimulus latency of 200-300 ms.