MEG Source Research Articles

Impact of different registration methods in MEG source analysis

Abstract For this study the impact of different co-registration procedures on MEG source localization of somatosensory evoked fields was evaluated. Two different co-registration procedures were used to calculate the transformation matrix which specifies how to align the MRI data to the MEG head coordinate system. In order to depict the differences, caused by the method, the Euclidian distance between the reconstructed sources was noted. It was shown that, erroneous MRI and MEG data co- registration effects source localization results. Most dipoles are located more posterior and superior when the more advanced registration procedure was applied. In conclusion the results show, that an iterative matching procedure allows an accurate knowledge of the MEG gradiometer sensor position relative to the head which is crucial to correctly reconstruct neuronal activity derived from MEG measurements.

MEG sensor and source measures of visually induced gamma-band oscillations are highly reliable

High frequency brain oscillations are associated with numerous cognitive and behavioral processes. Non-invasive measurements using electro-/magnetoencephalography (EEG/MEG) have revealed that high frequency neural signals are heritable and manifest changes with age as well as in neuropsychiatric illnesses. Despite the extensive use of EEG/MEG-measured neural oscillations in basic and clinical research, studies demonstrating test–retest reliability of power and frequency measures of neural signals remain scarce. Here, we evaluated the test–retest reliability of visually induced gamma (30–100Hz) oscillations derived from sensor and source signals acquired over two MEG sessions. The study required participants (N=13) to detect the randomly occurring stimulus acceleration while viewing a moving concentric grating. Sensor and source MEG measures of gamma-band activity yielded comparably strong reliability (average intraclass correlation, ICC=0.861). Peak stimulus-induced gamma frequency (53–72Hz) yielded the highest measures of stability (ICCsensor=0.940; ICCsource=0.966) followed by spectral signal change (ICCsensor=0.890; ICCsource=0.893) and peak frequency bandwidth (ICCsensor=0.856; ICCsource=0.622). Furthermore, source-reconstruction significantly improved signal-to-noise for spectral amplitude of gamma activity compared to sensor estimates. Our assessments highlight that both sensor and source derived estimates of visually induced gamma-band oscillations from MEG signals are characterized by high test–retest reliability, with source derived oscillatory measures conferring an improvement in the stability of peak-frequency estimates. Importantly, our finding of high test–retest reliability supports the feasibility of pharma-MEG studies and longitudinal aging or clinical studies.

Fully automated quality assurance and localization of volumetric MEG for single-subject mapping

BackgroundRobust and reproducible source mapping with magnetoencephalography is particularly challenging at the individual level. We evaluated a receiver-operating characteristic reliability (ROC-r) method for automated production of volumetric MEG maps in single-subjects. ROC-r provides quality assurance comparable to that offered by goodness-of-fit (GoF) and confidence volume (CV) for equivalent current dipole (ECD) modeling. New methodROC-r utilizes within-session reproducibility for quality assurance, latency identification, and thresholding of volumetric source maps. We tested ROC-r on simulated and real MEG with a strongly focal source, using somatosensory evoked fields (SEFs) elicited by bilateral median nerve stimulation (MNS). For quality assurance, the ROC-r reliable fraction (FR) was compared to the ECD GoF and CV. Peak beamformer locations and latencies identified by ROC-r were compared to the ECD for co-localization accuracy. ResultsThe predominant component of the SEF response occurred around 35ms, contralateral to the MNS. Comparison with existing methodsFR and 1/CV were more strongly correlated (mean Pearson’s correlation: 0.76; 95% CI 0.60–0.87) than FR and GoF (0.65; 95% CI 0.32–0.85). There was no difference in the latency of the peak GoF (35.0+/−0.6ms), CV (34.8+/−0.7ms) and FR (35.5+/−0.8ms). The ECD fits and ROC-r peaks co-localized to within a mean (median) distance of 8.3+/−5.9mm (6.2mm). ConclusionROC-r volumetric mapping co-localized closely with the standard ECD approach. This analysis can be added to any whole-brain MEG source imaging protocol, and is especially useful for single-subject mapping. Additionally, the development of FR as an analogue to GoF or CV for volumetric mapping is a critical improvement for clinical applications.

Spatiotemporal analysis of the appearance of gamma-band Microstates in resting state MEG.

Spatiotemporal analysis of EEG signal has revealed a rich set of methods to quantify neuronal activity using spatially global topographic templates, called Microstates. These methods complement more traditional spectral analysis, which uses band limited source data to determine defining differences in band power and peak characteristics. The high sampling rate and increased resistance to high frequency noise of MEG data offers an opportunity to explore the utility of spatiotemporal analysis over a wider spectrum than in EEG. In this work, we explore the utility of representing band limited MEG source data using established microstate techniques, especially in gamma frequency bands - a range yet unexplored using these techniques. We develop methods for gauging the goodness-of-fit achieved by resultant microstate templates and demonstrate sensor-level dispersion characteristics across wide-band signals as well as across signals filtered by canonical bands. These analyses reveal that, while high-frequency-band derived microstate templates are visually lawful, they fail to exhibit important explained variance and dispersion characteristics present in low- and full-band data necessary to meet the requirements of a microstate model.

Usefulness of multiple frequency band source localizations in ictal MEG

ObjectiveWe evaluated the diagnostic value of multiple frequency band MEG source localization within a wide time window during the preictal period. MethodsData for 13 epilepsy patients who showed an ictal event during MEG were analyzed. Several seconds of preictal data were localized in the theta, alpha, beta, and gamma bands by using wavelet transformation and the sLORETA algorithm. The same analysis was performed with narrow time and frequency band. Localization concordances to the surgically resected area were compared. ResultsSource localization in the gamma band for a 10s window before ictal onset showed best concordance to the resection cavity. Eight of 13 patients showed sub-lobar concordance in the 10s gamma band localization, whereas 3 showed concordance in the narrow time and frequency analysis. Four of 7 patients with focal cortical dysplasia (FCD) achieved seizure-free outcome, and all 4 showed sub-lobar concordance. ConclusionsA 10s time window gamma source localization method can be used to delineate the epileptogenic zone. SignificanceThe use of a long period during preictal gamma source localization has the potential to become a localizing biomarker of the epileptogenic zone in candidates for surgical intervention, especially in MRI-suspected FCD.

Modeling the effect of dendritic input location on MEG and EEG source dipoles.

The cerebral sources of magneto- and electroencephalography (MEG, EEG) signals can be represented by current dipoles. We used computational modeling of realistically shaped passive-membrane dendritic trees of pyramidal cells from the human cerebral cortex to examine how the spatial distribution of the synaptic inputs affects the current dipole. The magnitude of the total dipole moment vector was found to be proportional to the vertical location of the synaptic input. The dipole moment had opposite directions for inputs above and below a reversal point located near the soma. Inclusion of shunting-type inhibition either suppressed or enhanced the current dipole, depending on whether the excitatory and inhibitory synapses were on the same or opposite side of the reversal point. Relating the properties of the macroscopic current dipoles to dendritic current distributions can help to provide means for interpreting MEG and EEG data in terms of synaptic connection patterns within cortical areas.

Combined EEG/MEG Can Outperform Single Modality EEG or MEG Source Reconstruction in Presurgical Epilepsy Diagnosis

We investigated two important means for improving source reconstruction in presurgical epilepsy diagnosis. The first investigation is about the optimal choice of the number of epileptic spikes in averaging to (1) sufficiently reduce the noise bias for an accurate determination of the center of gravity of the epileptic activity and (2) still get an estimation of the extent of the irritative zone. The second study focuses on the differences in single modality EEG (80-electrodes) or MEG (275-gradiometers) and especially on the benefits of combined EEG/MEG (EMEG) source analysis. Both investigations were validated with simultaneous stereo-EEG (sEEG) (167-contacts) and low-density EEG (ldEEG) (21-electrodes). To account for the different sensitivity profiles of EEG and MEG, we constructed a six-compartment finite element head model with anisotropic white matter conductivity, and calibrated the skull conductivity via somatosensory evoked responses. Our results show that, unlike single modality EEG or MEG, combined EMEG uses the complementary information of both modalities and thereby allows accurate source reconstructions also at early instants in time (epileptic spike onset), i.e., time points with low SNR, which are not yet subject to propagation and thus supposed to be closer to the origin of the epileptic activity. EMEG is furthermore able to reveal the propagation pathway at later time points in agreement with sEEG, while EEG or MEG alone reconstructed only parts of it. Subaveraging provides important and accurate information about both the center of gravity and the extent of the epileptogenic tissue that neither single nor grand-averaged spike localizations can supply.

Influence of the head model on EEG and MEG source connectivity analyses

The results of brain connectivity analysis using reconstructed source time courses derived from EEG and MEG data depend on a number of algorithmic choices. While previous studies have investigated the influence of the choice of source estimation method or connectivity measure, the effects of the head modeling errors or simplifications have not been studied sufficiently.In the present simulation study, we investigated the influence of particular properties of the head model on the reconstructed source time courses as well as on source connectivity analysis in EEG and MEG. Therefore, we constructed a realistic head model and applied the finite element method to solve the EEG and MEG forward problems. We considered the distinction between white and gray matter, the distinction between compact and spongy bone, the inclusion of a cerebrospinal fluid (CSF) compartment, and the reduction to a simple 3-layer model comprising only the skin, skull, and brain. Source time courses were reconstructed using a beamforming approach and the source connectivity was estimated by the imaginary coherence (ICoh) and the generalized partial directed coherence (GPDC).Our results show that in both EEG and MEG, neglecting the white and gray matter distinction or the CSF causes considerable errors in reconstructed source time courses and connectivity analysis, while the distinction between spongy and compact bone is just of minor relevance, provided that an adequate skull conductivity value is used. Large inverse and connectivity errors are found in the same regions that show large topography errors in the forward solution. Moreover, we demonstrate that the very conservative ICoh is relatively safe from the crosstalk effects caused by imperfect head models, as opposed to the GPDC.

Direction of magnetoencephalography sources associated with feedback and feedforward contributions in a visual object recognition task

Identifying inter-area communication in terms of the hierarchical organization of functional brain areas is of considerable interest in human neuroimaging. Previous studies have suggested that the direction of magneto- and electroencephalography (MEG, EEG) source currents depend on the layer-specific input patterns into a cortical area. We examined the direction in MEG source currents in a visual object recognition experiment in which there were specific expectations of activation in the fusiform region being driven by either feedforward or feedback inputs. The source for the early non-specific visual evoked response, presumably corresponding to feedforward driven activity, pointed outward, i.e., away from the white matter. In contrast, the source for the later, object-recognition related signals, expected to be driven by feedback inputs, pointed inward, toward the white matter. Associating specific features of the MEG/EEG source waveforms to feedforward and feedback inputs could provide unique information about the activation patterns within hierarchically organized cortical areas.

MEG source reconstruction based on identification of directed source interactions on whole-brain anatomical networks

We present an MEG source reconstruction method that simultaneously reconstructs source amplitudes and identifies source interactions across the whole brain. In the proposed method, a full multivariate autoregressive (MAR) model formulates directed interactions (i.e., effective connectivity) between sources. The MAR coefficients (the entries of the MAR matrix) are constrained by the prior knowledge of whole-brain anatomical networks inferred from diffusion MRI. Moreover, to increase the accuracy and robustness of our method, we apply an fMRI prior on the spatial activity patterns and a sparse prior on the MAR coefficients. The observation process of MEG data, the source dynamics, and a series of the priors are combined into a Bayesian framework using a state-space representation. The parameters, such as the source amplitudes and the MAR coefficients, are jointly estimated from a variational Bayesian learning algorithm. By formulating the source dynamics in the context of MEG source reconstruction, and unifying the estimations of source amplitudes and interactions, we can identify the effective connectivity without requiring the selection of regions of interest. Our method is quantitatively and qualitatively evaluated on simulated and experimental data, respectively. Compared with non-dynamic methods, in which the interactions are estimated after source reconstruction with no dynamic constraints, the proposed dynamic method improves most of the performance measures in simulations, and provides better physiological interpretation and inter-subject consistency in real data applications.

Source-space ICA for EEG source separation, localization, and time-course reconstruction

We propose source-space independent component analysis (ICA) for separation, tomography, and time-course reconstruction of EEG and MEG source signals. Source-space ICA is based on the application of singular value decomposition and ICA on the neuroelectrical signals from all brain voxels obtained post minimum-variance beamforming of sensor-space EEG or MEG. We describe the theoretical background and equations, then evaluate the performance of this technique in several different situations, including weak sources, bilateral correlated sources, multiple sources, and cluster sources. In this approach, tomographic maps of sources are obtained by back-projection of the ICA mixing coefficients into the source-space (3-D brain template). The advantages of source-space ICA over the popular alternative approaches of sensor-space ICA together with dipole fitting and power mapping via minimum-variance beamforming are demonstrated. Simulated EEG data were produced by forward head modeling to project the simulated sources onto scalp sensors, then superimposed on real EEG background. To illustrate the application of source-space ICA to real EEG source reconstruction, we show the localization and time-course reconstruction of visual evoked potentials. Source-space ICA is superior to the minimum-variance beamforming in the reconstruction of multiple weak and strong sources, as ICA allows weak sources to be identified and reconstructed in the presence of stronger sources. Source-space ICA is also superior to sensor-space ICA on accuracy of localization of sources, as source-space ICA applies ICA to the time-courses of voxels reconstructed from minimum-variance beamforming on a 3D scanning grid and these time-courses are optimally unmixed via the beamformer. Each component identified by source-space ICA has its own tomographic map which shows the extent to which each voxel has contributed to that component.

Discrimination of cortical laminae using MEG

Typically MEG source reconstruction is used to estimate the distribution of current flow on a single anatomically derived cortical surface model. In this study we use two such models representing superficial and deep cortical laminae. We establish how well we can discriminate between these two different cortical layer models based on the same MEG data in the presence of different levels of co-registration noise, Signal-to-Noise Ratio (SNR) and cortical patch size. We demonstrate that it is possible to make a distinction between superficial and deep cortical laminae for levels of co-registration noise of less than 2mm translation and 2° rotation at SNR>11dB. We also show that an incorrect estimate of cortical patch size will tend to bias layer estimates. We then use a 3D printed head-cast (Troebinger et al., 2014) to achieve comparable levels of co-registration noise, in an auditory evoked response paradigm, and show that it is possible to discriminate between these cortical layer models in real data.

S15: Sensitivity and validity of MEG source localization in focal epilepsy: dependencies on source size and location

S15: Sensitivity and validity of MEG source localization in focal epilepsy: dependencies on source size and location

Combining EEG and MEG for the Reconstruction of Epileptic Activity Using a Calibrated Realistic Volume Conductor Model

To increase the reliability for the non-invasive determination of the irritative zone in presurgical epilepsy diagnosis, we introduce here a new experimental and methodological source analysis pipeline that combines the complementary information in EEG and MEG, and apply it to data from a patient, suffering from refractory focal epilepsy. Skull conductivity parameters in a six compartment finite element head model with brain anisotropy, constructed from individual MRI data, are estimated in a calibration procedure using somatosensory evoked potential (SEP) and field (SEF) data. These data are measured in a single run before acquisition of further runs of spontaneous epileptic activity. Our results show that even for single interictal spikes, volume conduction effects dominate over noise and need to be taken into account for accurate source analysis. While cerebrospinal fluid and brain anisotropy influence both modalities, only EEG is sensitive to skull conductivity and conductivity calibration significantly reduces the difference in especially depth localization of both modalities, emphasizing its importance for combining EEG and MEG source analysis. On the other hand, localization differences which are due to the distinct sensitivity profiles of EEG and MEG persist. In case of a moderate error in skull conductivity, combined source analysis results can still profit from the different sensitivity profiles of EEG and MEG to accurately determine location, orientation and strength of the underlying sources. On the other side, significant errors in skull modeling are reflected in EEG reconstruction errors and could reduce the goodness of fit to combined datasets. For combined EEG and MEG source analysis, we therefore recommend calibrating skull conductivity using additionally acquired SEP/SEF data.

Comparison of three-shell and simplified volume conductor models in magnetoencephalography

Experimental MEG source imaging studies have typically been carried out with either a spherically symmetric head model or a single-shell boundary-element (BEM) model that is shaped according to the inner skull surface. The concepts and comparisons behind these simplified models have led to misunderstandings regarding the role of skull and scalp in MEG. In this work, we assess the forward-model errors due to different skull/scalp approximations and due to differences and errors in model geometries.We built five anatomical models of a volunteer using a set of T1-weighted MR scans and three common toolboxes. Three of the models represented typical models in experimental MEG, one was manually constructed, and one contained a major segmentation error at the skull base. For these anatomical models, we built forward models using four simplified approaches and a three-shell BEM approach that has been used as reference in previous studies. Our reference model contained in addition the skull fine-structure (spongy bone).We computed signal topographies for cortically constrained sources in the left hemisphere and compared the topographies using relative error and correlation metrics. The results show that the spongy bone has a minimal effect on MEG topographies, and thus the skull approximation of the three-shell model is justified. The three-shell model performed best, followed by the corrected-sphere and single-shell models, whereas the local-spheres and single-sphere models were clearly worse. The three-shell model was the most robust against the introduced segmentation error. In contrast to earlier claims, there was no noteworthy difference in the computation times between the realistically-shaped and sphere-based models, and the manual effort of building a three-shell model and a simplified model is comparable. We thus recommend the realistically-shaped three-shell model for experimental MEG work. In cases where this is not possible, we recommend a realistically-shaped corrected-sphere or single-shell model.

Theta band activity in response to emotional expressions and its relationship with gamma band activity as revealed by MEG and advanced beamformer source imaging

Neuronal oscillations in the theta and gamma bands have been shown to be important for cognition. Here we examined the temporal and spatial relationship between the two frequency bands in emotional processing using magnetoencephalography and an advanced dynamic beamformer source imaging method called synthetic aperture magnetometry. We found that areas including the amygdala, visual and frontal cortex showed significant event-related synchronization in both bands, suggesting a functional association of neuronal oscillations in the same areas in the two bands. However, while the temporal profile in both bands was similar in the amygdala, the peak in gamma band power was much earlier within both visual and frontal areas. Our results do not support a traditional view that the localizations of lower and higher frequencies are spatially distinct. Instead, they suggest that in emotional processing, neuronal oscillations in the gamma and theta bands may reflect, at least in visual and frontal cortex either different but related functional processes or, perhaps more probably, different computational components of the same functional process.

Two‐way regularization for MEG source reconstruction via multilevel coordinate descent

AbstractMagnetoencephalography (MEG) source reconstruction refers to the inverse problem of recovering the neural activity from the MEG time course measurements. A spatiotemporal two‐way regularization (TWR) method was recently proposed by Tian et al. to solve this inverse problem and was shown to outperform several one‐way regularization methods and spatiotemporal methods. This TWR method is a two‐stage procedure that first obtains a raw estimate of the source signals and then refines the raw estimate to ensure spatial focality and temporal smoothness using spatiotemporal regularized matrix decomposition. Although proven to be effective, the performance of two‐stage TWR depends on the quality of the raw estimate. In this paper we directly solve the MEG source reconstruction problem using a multivariate penalized regression where the number of variables is much larger than the number of cases. A special feature of this regression is that the regression coefficient matrix has a spatiotemporal two‐way structure that naturally invites a two‐way penalty. Making use of this structure, we develop a computationally efficient multilevel coordinate descent algorithm to implement the method. This new one‐stage TWR method has shown its superiority to the two‐stage TWR method in three simulation studies with different levels of complexity and a real‐world MEG data analysis. © 2013 Wiley Periodicals, Inc. Statistical Analysis and Data Mining, 2013

Reliability of MEG source imaging of anterior temporal spikes: Analysis of an intracranially characterized spike focus

ObjectiveTo assess the reliability of MEG source imaging (MSI) of anterior temporal spikes through detailed analysis of the localization and orientation of source solutions obtained for a large number of spikes that were separately confirmed by intracranial EEG to be focally generated within a single, well-characterized spike focus. MethodsMSI was performed on 64 identical right anterior temporal spikes from an anterolateral temporal neocortical spike focus. The effects of different volume conductors (sphere and realistic head model), removal of noise with low frequency filters (LFFs) and averaging multiple spikes were assessed in terms of the reliability of the source solutions. ResultsMSI of single spikes resulted in scattered dipole source solutions that showed reasonable reliability for localization at the lobar level, but only for solutions with a goodness-of-fit exceeding 80% using a LFF of 3Hz. Reliability at a finer level of intralobar localization was limited. Spike averaging significantly improved the reliability of source solutions and averaging 8 or more spikes reduced dependency on goodness-of-fit and data filtering. ConclusionsMSI performed on topographically identical individual spikes from an intracranially defined classical anterior temporal lobe spike focus was limited by low reliability (i.e., scattered source solutions) in terms of fine, sublobar localization within the ipsilateral temporal lobe. Spike averaging significantly improved reliability. SignificanceMSI performed on individual anterior temporal spikes is limited by low reliability. Reduction of background noise through spike averaging significantly improves the reliability of MSI solutions.

Reconstructing spatially extended brain sources via enforcing multiple transform sparseness

Accurate estimation of location and extent of neuronal sources from EEG/MEG remain challenging. In the present study, a new source imaging method, i.e. variation and wavelet based sparse source imaging (VW-SSI), is proposed to better estimate cortical source locations and extents. VW-SSI utilizes the L1-norm regularization method with the enforcement of transform sparseness in both variation and wavelet domains. The performance of the proposed method is assessed by both simulated and experimental MEG data, obtained from a language task and a motor task. Compared to L2-norm regularizations, VW-SSI demonstrates significantly improved capability in reconstructing multiple extended cortical sources with less spatial blurredness and less localization error. With the use of transform sparseness, VW-SSI overcomes the over-focused problem in classic SSI methods. With the use of two transformations, VW-SSI further indicates significantly better performance in estimating MEG source locations and extents than other SSI methods with single transformations. The present experimental results indicate that VW-SSI can successfully estimate neural sources (and their spatial coverage) located in close areas while responsible for different functions, i.e. temporal cortical sources for auditory and language processing, and sources on the pre-bank and post-bank of the central sulcus. Meantime, all other methods investigated in the present study fail to recover these phenomena. Precise estimation of cortical source locations and extents from EEG/MEG is of significance for applications in neuroscience and neurology.

MEG and EEG Sensitivity in a Case of Medial Occipital Epilepsy

Interictal or ictal events in partial epilepsies may project on scalp EEG contralaterally to the side of the epileptogenic lesion. Such paradoxical lateralization can be observed in case of para-sagittal generators, and is likely due to the spatial orientation of the generator, presenting an oblique projection towards the midline. We present here a case of medial occipital epilepsy investigated using EEG, MEG and stereoelectroencephalography (SEEG). MRI displayed a focal cortical dysplasia in the superior margin of the right calcarine fissure. SEEG demonstrated bilateral medial occipital interictal spikes, with an inversion of polarity at the level of the lesion and a contralateral propagation occurring in 10ms. Interictal iterative EEG cartographies showed a large posterior field, with a maximum contralateral to the initial generator (EEG paradoxical lateralization). With the same number of channels, interictal iterative MEG cartographies were more precise and more complex than EEG ones, indicating an onset accurately lateralized. A few milliseconds later, MEG cartographies were quadripolar, thus indicating two homotopic active generators. These MEG and EEG cartographies have been reproduced using BESA dipole simulator. Relative merits of MEG and EEG are still debated. With 151 channels, MEG source localizations indicated the right medial occipital area, as demonstrated by SEEG. An investigation with a corresponding number of EEG channels was not performed. After a down sampling to 64 sensors, this precision was lost. MEG and EEG source localization results, both with 64 channels, were quite comparable, indicating both medial occipital areas. However, a careful analysis of MEG/EEG iterative cartographies, performed with the same number of channels in both modalities, demonstrated that, in this configuration, MEG sensitivity was superior to the EEG one, allowing separating two medial occipital sources, characterized in SEEG by a time delay of 10ms.