Magnetoencephalography can reveal deep brain network activities linked to memory processes

Recording from deep neural structures such as hippocampus noninvasively and yet with high temporal resolution remains a major challenge for human neuroscience. Although it has been proposed that deep neuronal activity might be recordable during cognitive tasks using magnetoencephalography (MEG), this remains to be demonstrated as the contribution of deep structures to MEG recordings may be too small to be detected or might be eclipsed by the activity of large‐scale neocortical networks. In the present study, we disentangled mesial activity and large‐scale networks from the MEG signals thanks to blind source separation (BSS). We then validated the MEG BSS components using intracerebral EEG signals recorded simultaneously in patients during their presurgical evaluation of epilepsy. In the MEG signals obtained during a memory task involving the recognition of old and new images, we identified with BSS a putative mesial component, which was present in all patients and all control subjects. The time course of the component selectively correlated with stereo‐electroencephalography signals recorded from hippocampus and rhinal cortex, thus confirming its mesial origin. This finding complements previous studies with epileptic activity and opens new possibilities for using MEG to study deep brain structures in cognition and in brain disorders.     

High‐precision magnetoencephalography for reconstructing amygdalar and hippocampal oscillations during prediction of safety and threat

Learning to associate neutral with aversive events in rodents is thought to depend on hippocampal and amygdala oscillations. In humans, oscillations underlying aversive learning are not well characterised, largely due to the technical difficulty of recording from these two structures. Here, we used high‐precision magnetoencephalography (MEG) during human discriminant delay threat conditioning. We constructed generative anatomical models relating neural activity with recorded magnetic fields at the single‐participant level, including the neocortex with or without the possibility of sources originating in the hippocampal and amygdalar structures. Models including neural activity in amygdala and hippocampus explained MEG data during threat conditioning better than exclusively neocortical models. We found that in both amygdala and hippocampus, theta oscillations during anticipation of an aversive event had lower power compared to safety, both during retrieval and extinction of aversive memories. At the same time, theta synchronisation between hippocampus and amygdala increased over repeated retrieval of aversive predictions, but not during safety. Our results suggest that high‐precision MEG is sensitive to neural activity of the human amygdala and hippocampus during threat conditioning and shed light on the oscillation‐mediated mechanisms underpinning retrieval and extinction of fear memories in humans.     

Detection and localization of hippocampal activity using beamformers with MEG: a detailed investigation using simulations and empirical data

The ability to detect neuronal activity emanating from deep brain structures such as the hippocampus using magnetoencephalography has been debated in the literature. While a significant number of recent publications reported activations from deep brain structures, others reported their inability to detect such activity even when other detection modalities confirmed its presence. In this article, we relied on realistic simulations to show that both sides of this debate are correct and that these findings are reconcilable. We show that the ability to detect such activations in evoked responses depends on the signal strength, the amount of brain noise background, the experimental design parameters, and the methodology used to detect them. Furthermore, we show that small signal strengths require contrasts with control conditions to be detected, particularly in the presence of strong brain noise backgrounds. We focus on one localization technique, the adaptive spatial filter (beamformer), and examine its strengths and weaknesses in reconstructing hippocampal activations, in the presence of other strong brain sources such as visual activations, and compare the performance of the vector and scalar beamformers under such conditions. We show that although a weight-normalized beamformer combined with a multisphere head model is not biased in the presence of uncorrelated random noise, it can be significantly biased in the presence of correlated brain noise. Furthermore, we show that the vector beamformer performs significantly better than the scalar under such conditions. We corroborate our findings empirically using real data and demonstrate our ability to detect and localize such sources.     

Magnetoencephalography as a research tool in neuroscience: state of the art.

Magnetoencephalography (MEG) is a noninvasive neuroimaging method for detecting, analyzing, and interpreting the magnetic field generated by the electrical activity in the brain. Modern hardware can capture the MEG signal at hundreds of points around the head in a snapshot lasting only a fraction of a millisecond. The sensitivity of modern hardware is high enough to permit the extraction of a clean signal generated by the brain well above the noise level of the MEG hardware. It is possible to identify signatures of superficial and often deep generators in the raw MEG signal, even in snapshots of data. In a more quantitative way, tomographic images of the electrical current density in the brain can be extracted from each snapshot of MEG signal, providing a direct correlate of coherent collective neuronal activity. A number of recent studies have scrutinized brain function in the new spatiotemporal window that real-time tomographic analysis of MEG signals has opened. The results have allowed the variability in a single area to be seen in the context of activity in other areas and background rhythmic activity. In this view, normal brain function is seen as a cascade of extremely fast events and the unfolding of specialized processes, segregated in space and time and organized into well-defined stages of processing.     

Enhanced brain signal variability in children with autism spectrum disorder during early childhood

Extensive evidence shows that a core neurobiological mechanism of autism spectrum disorder (ASD) involves aberrant neural connectivity. Recent advances in the investigation of brain signal variability have yielded important information about neural network mechanisms. That information has been applied fruitfully to the assessment of aging and mental disorders. Multiscale entropy (MSE) analysis can characterize the complexity inherent in brain signal dynamics over multiple temporal scales in the dynamics of neural networks. For this investigation, we sought to characterize the magnetoencephalography (MEG) signal variability during free watching of videos without sound using MSE in 43 children with ASD and 72 typically developing controls (TD), emphasizing early childhood to older childhood: a critical period of neural network maturation. Results revealed an age‐related increase of brain signal variability in a specific timescale in TD children, whereas atypical age‐related alteration was observed in the ASD group. Additionally, enhanced brain signal variability was observed in children with ASD, and was confirmed particularly for younger children. In the ASD group, symptom severity was associated region‐specifically and timescale‐specifically with reduced brain signal variability. These results agree well with a recently reported theory of increased brain signal variability during development and aberrant neural connectivity in ASD, especially during early childhood. Results of this study suggest that MSE analytic method might serve as a useful approach for characterizing neurophysiological mechanisms of typical‐developing and its alterations in ASD through the detection of MEG signal variability at multiple timescales. Hum Brain Mapp 37:1038–1050, 2016. © 2015 The Authors Human Brain Mapping Published by Wiley Periodicals, Inc .     

Modulation of hippocampal theta and hippocampal‐prefrontal cortex function by a schizophrenia risk gene

Hippocampal theta‐band oscillations are thought to facilitate the co‐ordination of brain activity across distributed networks, including between the hippocampus and prefrontal cortex (PFC). Impairments in hippocampus‐PFC functional connectivity are implicated in schizophrenia and are associated with a polymorphism within the ZNF804A gene that shows a genome‐wide significant association with schizophrenia. However, the mechanisms by which ZNF804A affects hippocampus‐PFC connectivity are unknown. We used a multimodal imaging approach to investigate the impact of the ZNF804A polymorphism on hippocampal theta and hippocampal network coactivity. Healthy volunteers homozygous for the ZNF804A rs1344706 (A[risk]/C[nonrisk]) polymorphism were imaged at rest using both magnetoencephalography (MEG) and functional magnetic resonance imaging (fMRI). A dual‐regression approach was used to investigate coactivations between the hippocampal network and other brain regions for both modalities, focusing on the theta band in the case of MEG. We found a significant decrease in intrahippocampal theta (using MEG) and greater coactivation of the superior frontal gyrus with the hippocampal network (using fMRI) in risk versus nonrisk homozygotes. Furthermore, these measures showed a significant negative correlation. Our demonstration of an inverse relationship between hippocampal theta and hippocampus‐PFC coactivation supports a role for hippocampal theta in coordinating hippocampal‐prefrontal activity. The ZNF804A‐related differences that we find in hippocampus‐PFC coactivation are consistent with previously reported associations with functional connectivity and with these changes lying downstream of altered hippocampal theta. Changes in hippocampal‐PFC co‐ordination, driven by differences in oscillatory activity, may be one mechanism by which ZNF804A impacts on brain function and risk for psychosis. Hum Brain Mapp 36:2387–2395, 2015. © 2015 The Authors Human Brain Mapping Published by Wiley Periodicals, Inc.     

A framework for the design of flexible cross‐talk functions for spatial filtering of EEG/MEG data: DeFleCT

Brain activation estimated from EEG and MEG data is the basis for a number of time‐series analyses. In these applications, it is essential to minimize “leakage” or “cross‐talk” of the estimates among brain areas. Here, we present a novel framework that allows the design of flexible cross‐talk functions (DeFleCT), combining three types of constraints: (1) full separation of multiple discrete brain sources, (2) minimization of contributions from other (distributed) brain sources, and (3) minimization of the contribution from measurement noise. Our framework allows the design of novel estimators by combining knowledge about discrete sources with constraints on distributed source activity and knowledge about noise covariance. These estimators will be useful in situations where assumptions about sources of interest need to be combined with uncertain information about additional sources that may contaminate the signal (e.g. distributed sources), and for which existing methods may not yield optimal solutions. We also show how existing estimators, such as maximum‐likelihood dipole estimation, L2 minimum‐norm estimation, and linearly‐constrained minimum variance as well as null‐beamformers, can be derived as special cases from this general formalism. The performance of the resulting estimators is demonstrated for the estimation of discrete sources and regions‐of‐interest in simulations of combined EEG/MEG data. Our framework will be useful for EEG/MEG studies applying time‐series analysis in source space as well as for the evaluation and comparison of linear estimators. Hum Brain Mapp 35:1642–1653, 2014. © 2013 Wiley Periodicals, Inc.     

Simultaneous EEG and MEG source reconstruction in sparse electromagnetic source imaging

Electroencephalography (EEG) and magnetoencephalography (MEG) have different sensitivities to differently configured brain activations, making them complimentary in providing independent information for better detection and inverse reconstruction of brain sources. In the present study, we developed an integrative approach, which integrates a novel sparse electromagnetic source imaging method, i.e., variation‐based cortical current density (VB‐SCCD), together with the combined use of EEG and MEG data in reconstructing complex brain activity. To perform simultaneous analysis of multimodal data, we proposed to normalize EEG and MEG signals according to their individual noise levels to create unit‐free measures. Our Monte Carlo simulations demonstrated that this integrative approach is capable of reconstructing complex cortical brain activations (up to 10 simultaneously activated and randomly located sources). Results from experimental data showed that complex brain activations evoked in a face recognition task were successfully reconstructed using the integrative approach, which were consistent with other research findings and validated by independent data from functional magnetic resonance imaging using the same stimulus protocol. Reconstructed cortical brain activations from both simulations and experimental data provided precise source localizations as well as accurate spatial extents of localized sources. In comparison with studies using EEG or MEG alone, the performance of cortical source reconstructions using combined EEG and MEG was significantly improved. We demonstrated that this new sparse ESI methodology with integrated analysis of EEG and MEG data could accurately probe spatiotemporal processes of complex human brain activations. This is promising for noninvasively studying large‐scale brain networks of high clinical and scientific significance. Hum Brain Mapp, 2013. © 2010 Wiley Periodicals, Inc.     

Magnetoencephalography in the study of children with autism spectrum disorder

Magnetoencephalography (MEG) is a non‐invasive neuroimaging technique that provides a measure of cortical neural activity on a millisecond timescale with high spatial resolution. MEG has been clinically applied to various neurological diseases, including epilepsy and cognitive dysfunction. In the past decade, MEG has also emerged as an important investigatory tool in neurodevelopmental studies. It is therefore an opportune time to review how MEG is able to contribute to the study of atypical brain development. We limit this review to autism spectrum disorder (ASD). The relevant published work for children was accessed using PubMed on 5 January 2015. Case reports, case series, and papers on epilepsy were excluded. Owing to their accurate separation of brain activity in the right and left hemispheres and the higher accuracy of source localization, MEG studies have added new information related to auditory‐evoked brain responses to findings from previous electroencephalography studies of children with ASD. In addition, evidence of atypical brain connectivity in children with ASD has accumulated over the past decade. MEG is well suited for the study of neural activity with high time resolution even in young children. Although further studies are still necessary, the detailed findings provided by neuroimaging methods may aid clinical diagnosis and even contribute to the refinement of diagnostic categories for neurodevelopmental disorders in the future.     

Predicting EEG responses using MEG sources in superior temporal gyrus reveals source asynchrony in patients with schizophrenia.

OBJECTIVE: An integrated analysis using Electroencephalography (EEG) and magnetoencephalography (MEG) is introduced to study abnormalities in early cortical responses to auditory stimuli in schizophrenia. METHODS: Auditory responses were recorded simultaneously using EEG and MEG from 20 patients with schizophrenia and 19 control subjects. Bilateral superior temporal gyrus (STG) sources and their time courses were obtained using MEG for the 30-100 ms post-stimulus interval. The MEG STG source time courses were used to predict the EEG signal at electrode Cz. RESULTS: In control subjects, the STG sources predicted the EEG Cz recording very well (97% variance explained). In schizophrenia patients, the STG sources accounted for substantially (86%) and significantly (P<0.0002) less variance. After MEG-derived STG activity was removed from the EEG Cz signal, the residual signal was dominated by 40 Hz activity, an indication that the remaining variance in EEG is probably contributed by other brain generators, rather than by random noise. CONCLUSIONS: Integrated MEG and EEG analysis can differentiate patients and controls, and suggests a basis for a well established abnormality in the cortical auditory response in schizophrenia, implicating a disorder of functional connectivity in the relationship between STG sources and other brain generators.     

Detectability of cerebellar activity with magnetoencephalography and electroencephalography

Electrophysiological signals from the cerebellum have traditionally been viewed as inaccessible to magnetoencephalography (MEG) and electroencephalography (EEG). Here, we challenge this position by investigating the ability of MEG and EEG to detect cerebellar activity using a model that employs a high‐resolution tessellation of the cerebellar cortex. The tessellation was constructed from repetitive high‐field (9.4T) structural magnetic resonance imaging (MRI) of an ex vivo human cerebellum. A boundary‐element forward model was then used to simulate the M/EEG signals resulting from neural activity in the cerebellar cortex. Despite significant signal cancelation due to the highly convoluted cerebellar cortex, we found that the cerebellar signal was on average only 30–60% weaker than the cortical signal. We also made detailed M/EEG sensitivity maps and found that MEG and EEG have highly complementary sensitivity distributions over the cerebellar cortex. Based on previous fMRI studies combined with our M/EEG sensitivity maps, we discuss experimental paradigms that are likely to offer high M/EEG sensitivity to cerebellar activity. Taken together, these results show that cerebellar activity should be clearly detectable by current M/EEG systems with an appropriate experimental setup.     

Three brain states in the hippocampus and cortex

Contemporary brain research seeks to understand how cognition is reducible to neural activity. Crucially, much of this effort is guided by a scientific paradigm that views neural activity as essentially driven by external stimuli. In contrast, recent perspectives argue that this paradigm is by itself inadequate and that understanding patterns of activity intrinsic to the brain is needed to explain cognition. Yet, despite this critique, the stimulus‐driven paradigm still dominates—possibly because a convincing alternative has not been clear. Here, we review a series of findings suggesting such an alternative. These findings indicate that neural activity in the hippocampus occurs in one of three brain states that have radically different anatomical, physiological, representational, and behavioral correlates, together implying different functional roles in cognition. This three‐state framework also indicates that neural representations in the hippocampus follow a surprising pattern of organization at the timescale of ～1 s or longer. Lastly, beyond the hippocampus, recent breakthroughs indicate three parallel states in the cortex, suggesting shared principles and brain‐wide organization of intrinsic neural activity.     

Distinct Neurophysiological Correlates of the fMRI BOLD Signal in the Hippocampus and Neocortex

Functional magnetic resonance imaging (fMRI) is among the foremost methods for mapping human brain function but provides only an indirect measure of underlying neural activity. Recent findings suggest that the neurophysiological correlates of the fMRI blood oxygenation level-dependent (BOLD) signal might be regionally specific. We examined the neurophysiological correlates of the fMRI BOLD signal in the hippocampus and neocortex, where differences in neural architecture might result in a different relationship between the respective signals. Fifteen human neurosurgical patients (10 female, 5 male) implanted with depth electrodes performed a verbal free recall task while electrophysiological activity was recorded simultaneously from hippocampal and neocortical sites. The same patients subsequently performed a similar version of the task during a later fMRI session. Subsequent memory effects (SMEs) were computed for both imaging modalities as patterns of encoding-related brain activity predictive of later free recall. Linear mixed-effects modeling revealed that the relationship between BOLD and gamma-band SMEs was moderated by the lobar location of the recording site. BOLD and high gamma (70–150 Hz) SMEs positively covaried across much of the neocortex. This relationship was reversed in the hippocampus, where a negative correlation between BOLD and high gamma SMEs was evident. We also observed a negative relationship between BOLD and low gamma (30–70 Hz) SMEs in the medial temporal lobe more broadly. These results suggest that the neurophysiological correlates of the BOLD signal in the hippocampus differ from those observed in the neocortex.SIGNIFICANCE STATEMENT The BOLD signal forms the basis of fMRI but provides only an indirect measure of neural activity. Task-related modulation of BOLD signals are typically equated with changes in gamma-band activity; however, relevant empirical evidence comes largely from the neocortex. We examined neurophysiological correlates of the BOLD signal in the hippocampus, where the differing neural architecture might result in a different relationship between the respective signals. We identified a positive relationship between encoding-related changes in BOLD and gamma-band activity in the frontal and parietal cortices. This effect was reversed in the hippocampus, where BOLD and gamma-band effects negatively covaried. These results suggest regional variability in the transfer function between neural activity and the BOLD signal in the hippocampus and neocortex.     

Potential of on‐scalp MEG: Robust detection of human visual gamma‐band responses

Electrophysiological signals recorded intracranially show rich frequency content spanning from near‐DC to hundreds of hertz. Noninvasive electromagnetic signals measured with electroencephalography (EEG) or magnetoencephalography (MEG) typically contain less signal power in high frequencies than invasive recordings. Particularly, noninvasive detection of gamma‐band activity (>30 Hz) is challenging since coherently active source areas are small at such frequencies and the available imaging methods have limited spatial resolution. Compared to EEG and conventional SQUID‐based MEG, on‐scalp MEG should provide substantially improved spatial resolution, making it an attractive method for detecting gamma‐band activity. Using an on‐scalp array comprised of eight optically pumped magnetometers (OPMs) and a conventional whole‐head SQUID array, we measured responses to a dynamic visual stimulus known to elicit strong gamma‐band responses. OPMs had substantially higher signal power than SQUIDs, and had a slightly larger relative gamma‐power increase over the baseline. With only eight OPMs, we could obtain gamma‐activity source estimates comparable to those of SQUIDs at the group level. Our results show the feasibility of OPMs to measure gamma‐band activity. To further facilitate the noninvasive detection of gamma‐band activity, the on‐scalp OPM arrays should be optimized with respect to sensor noise, the number of sensors and intersensor spacing.     

Regulation of gamma‐frequency oscillation by feedforward inhibition: A computational modeling study

Throughout the brain, reciprocally connected excitatory and inhibitory neurons interact to produce gamma‐frequency oscillations. The emergent gamma rhythm synchronizes local neural activity and helps to select which cells should fire in each cycle. We previously found that such excitation‐inhibition microcircuits, however, have a potentially significant caveat: the frequency of the gamma oscillation and the level of selection (i.e., the percentage of cells that are allowed to fire) vary with the magnitude of the input signal. In networks with varying levels of brain activity, such a feature may produce undesirable instability on the time and spatial structure of the neural signal with a potential for adversely impacting important neural processing mechanisms. Here we propose that feedforward inhibition solves the latter instability problem of the excitation‐inhibition microcircuit. Using computer simulations, we show that the feedforward inhibitory signal reduces the dependence of both the frequency of population oscillation and the level of selection on the magnitude of the input excitation. Such a mechanism can produce stable gamma oscillations with its frequency determined only by the properties of the feedforward network, as observed in the hippocampus. As feedforward and feedback inhibition motifs commonly appear together in the brain, we hypothesize that their interaction underlies a robust implementation of general computational principles of neural processing involved in several cognitive tasks, including the formation of cell assemblies and the routing of information between brain areas.     

Spectral graph theory of brain oscillations

The relationship between the brain's structural wiring and the functional patterns of neural activity is of fundamental interest in computational neuroscience. We examine a hierarchical, linear graph spectral model of brain activity at mesoscopic and macroscopic scales. The model formulation yields an elegant closed‐form solution for the structure–function problem, specified by the graph spectrum of the structural connectome's Laplacian, with simple, universal rules of dynamics specified by a minimal set of global parameters. The resulting parsimonious and analytical solution stands in contrast to complex numerical simulations of high dimensional coupled nonlinear neural field models. This spectral graph model accurately predicts spatial and spectral features of neural oscillatory activity across the brain and was successful in simultaneously reproducing empirically observed spatial and spectral patterns of alpha‐band (8–12 Hz) and beta‐band (15–30 Hz) activity estimated from source localized magnetoencephalography (MEG). This spectral graph model demonstrates that certain brain oscillations are emergent properties of the graph structure of the structural connectome and provides important insights towards understanding the fundamental relationship between network topology and macroscopic whole‐brain dynamics. .     

Head models and dynamic causal modeling of subcortical activity using magnetoencephalographic/electroencephalographic data

Cognitive functions involve not only cortical but also subcortical structures. Subcortical sources, however, contribute very little to magnetoencephalographic (MEG) and electroencephalographic (EEG) signals because they are far from external sensors and their neural architectonic organization often makes them electromagnetically silent. Estimating the activity of deep sources from MEG and EEG (M/EEG) data is thus a challenging issue. Here, we review the influence of geometric parameters (location/orientation) on M/EEG signals produced by the main deep brain structures (amygdalo-hippocampal complex, thalamus and some basal ganglia). We then discuss several methods that have been utilized to solve the issues and localize or quantify the M/EEG contribution from deep neural currents. These methods rely on realistic forward models of subcortical regions or on introducing strong dynamical priors on inverse solutions that are based on biologically plausible neural models, such as those used in dynamic causal modeling (DCM) for M/EEG.     

Functional and structural correlates of the aging brain: Relating visual cortex (V1) gamma band responses to age‐related structural change

The gamma band response is thought to be a key neural signature of information processing in the mammalian brain, yet little is known about how age‐related maturation influences the γ‐band response. Recent MRI‐based studies have shown that brain maturation is accompanied by clear structural changes in both gray and white matter, yet the correspondence of these changes to brain function is unclear. The objective of this study was to relate visual cortex (V1) γ‐band responses to age‐related structural change. We evaluated MEG measured γ‐band responses to contrast gratings stimuli and structural MRIs from participants observed from two separate research centers (MEG lab at CUBRIC, Cardiff University, UK, and the Lurie Family Foundations MEG Imaging Center, (CHOP) at the Children's Hospital of Philadelphia). Pooled participant data (N = 59) ranged in age from 8.7 to 45.3 years. We assessed linear associations between age and MEG γ‐band frequency and amplitude, as well as between age and MRI volumetric parameters of the occipital lobe. Our MEG findings revealed a significant negative correlation for gamma band frequency versus age. Volumetric brain analysis from the occipital lobe also revealed significant negative correlations between age and the cortical thickness of pericalcarine and cuneus areas. Our functional MEG and structural MRI findings shows regionally specific changes due to maturation and may thus be informative for understanding physiological processes of neural development, maturation, and age‐related decline. In addition, this study represents (to our knowledge), the first published demonstration of multicenter data sharing across MEG centers. Hum Brain Mapp 33:2035–2046, 2012. © 2011 Wiley Periodicals, Inc.     

Identifying fragments of natural speech from the listener's MEG signals

It is a challenge for current signal analysis approaches to identify the electrophysiological brain signatures of continuous natural speech that the subject is listening to. To relate magnetoencephalographic (MEG) brain responses to the physical properties of such speech stimuli, we applied canonical correlation analysis (CCA) and a Bayesian mixture of CCA analyzers to extract MEG features related to the speech envelope. Seven healthy adults listened to news for an hour while their brain signals were recorded with whole‐scalp MEG. We found shared signal time series (canonical variates) between the MEG signals and speech envelopes at 0.5–12 Hz. By splitting the test signals into equal‐length fragments from 2 to 65 s (corresponding to 703 down to 21 pieces per the total speech stimulus) we obtained better than chance‐level identification for speech fragments longer than 2–3 s, not used in the model training. The applied analysis approach thus allowed identification of segments of natural speech by means of partial reconstruction of the continuous speech envelope (i.e., the intensity variations of the speech sounds) from MEG responses, provided means to empirically assess the time scales obtainable in speech decoding with the canonical variates, and it demonstrated accurate identification of the heard speech fragments from the MEG data. Hum Brain Mapp, 2013. © 2012 Wiley Periodicals, Inc.     

Feasibility and limitations of magnetoencephalographic detection of epileptic discharges: simultaneous recording of magnetic fields and electrocorticography

Magnetoencephalography (MEG) is considered clinically useful in localizing the epileptogenic focus in partial epilepsy. However, the relationship between the extent of the brain involved in paroxysmal activities and the magnetic field changes at the scalp has not been fully clarified. Furthermore, whether paroxysmal activities generated in deep brain structures such as the hippocampus can be detected magnetically is uncertain. Eight patients with temporal lobe epilepsy and two with extratemporal lobe epilepsy underwent chronic recording from subdural electrodes. Magnetic and electrocorticographic discharges representing epileptic activity were recorded simultaneously. MEG recorded magnetic field changes originating from paroxysmal activity in the superiolateral cerebral cortex when the amplitudes of the electrical paroxysmal activities exceeded 100 microV and extended over more than 3 cm2 of cortical surface. MEG failed to record paroxysmal activity localized to the medial temporal lobe. MEG is often useful in identifying a spike focus in the superiolateral aspects of the cerebral hemisphere, but not discharges arising from the medial temporal lobe. Rapid decay of the magnetic field is likely to be the reason for this limited sensitivity to medial discharges.