Non-invasive long-term recordings of cortical 'direct current' (DC-) activity in humans using magnetoencephalography

Updated review on the link between cortical spreading depression and headache disorders

INTRODUCTION

Experimental animal studies have revealed mechanisms that link cortical spreading depression (CSD) to the trigeminal activation mediating lateralized headache. However, conventional CSD as seen in lissencephalic brain is insufficient to explain some clinical features of aura and migraine headache.

AREAS COVERED

The importance of CSD in headache development including dysfunction of the thalamocortical network, neuroinflammation, calcitonin gene-related peptide, transgenic models, and the role of CSD in migraine triggers, treatment options, neuromodulation and future directions are reviewed.

EXPERT OPINION

The conventional understanding of CSD marching across the hemisphere is invalid in gyrencephalic brains. Thalamocortical dysfunction and interruption of functional cortical network systems by CSD, may provide alternative explanations for clinical manifestations of migraine phases including aura. Not all drugs showing CSD blocking properties in lissencephalic brains, have efficacy in migraine headache and monoclonal antibodies against CGRP ligand/receptors which are effective in migraine treatment, have no impact on aura in humans or CSD properties in rodents. Functional networks and molecular mechanisms mediating and amplifying the effects of limited CSD in migraine brain remain to be investigated to define new targets.

Aura and Head pain: relationship and gaps in the translational models

Migraine is a complex brain disorder and initiating events for acute attacks still remain unclear. It seems difficult to explain the development of migraine headache with one mechanism and/or a single anatomical location. Cortical spreading depression (CSD) is recognized as the biological substrate of migraine aura and experimental animal studies have provided mechanisms that possibly link CSD to the activation of trigeminal neurons mediating lateralized head pain. However, some CSD features do not match the clinical features of migraine headache and there are gaps in translating CSD to migraine with aura. Clinical features of migraine headache and results from research are critically evaluated; and consistent and inconsistent findings are discussed according to the known basic features of canonical CSD: typical SD limited to the cerebral cortex as it was originally defined. Alternatively, arguments related to the emergence of SD in other brain structures in addition to the cerebral cortex or CSD initiated dysfunction in the thalamocortical network are proposed. Accordingly, including thalamus, particularly reticular nucleus and higher order thalamic nuclei, which functions as a hub connecting the visual, somatosensory, language and motor cortical areas and subjects to modulation by brain stem projections into the CSD theory, would greatly improve our current understanding of migraine.

Sound processing hierarchy within human auditory cortex

Both attention and masking sounds can alter auditory neural processes and affect auditory signal perception. In the present study, we investigated the complex effects of auditory-focused attention and the signal-to-noise ratio of sound stimuli on three different auditory evoked field components (auditory steady-state response, N1m, and sustained field) by means of magnetoencephalography. The results indicate that the auditory steady-state response originating in primary auditory cortex reflects the signal-to-noise ratio of physical sound inputs (bottom-up process) rather than the listener's attentional state (top-down process), whereas the sustained field, originating in nonprimary auditory cortex, reflects the attentional state rather than the signal-to-noise ratio. The N1m was substantially influenced by both bottom-up and top-down neural processes. The differential sensitivity of the components to bottom-up and top-down neural processes, contingent on their level in the processing pathway, suggests a stream from bottom-up driven sensory neural processing to top-down driven auditory perception within human auditory cortex.

Differential infraslow (<0.1 Hz) cortical activations in the affected and unaffected hemispheres from patients with subacute stroke demonstrated by noninvasive DC-magnetoencephalography

BACKGROUND AND PURPOSE

Sustained mass depolarization of neurons, termed cortical spreading depolarization, is one electrophysiological correlate of the ischemic injury of neurons. Cortical spreading depolarizations spread in the gray matter at a rate of approximately 3 mm/min and are associated with large infraslow extracellular potential changes (<0.05 Hz). Moreover, smaller infraslow potential changes accompany functional activation and might help to assess neuronal repair after stroke. The objective of the present pilot study was to investigate whether it is feasible to apply noninvasive near-DC-magnetoencephalography to detect and monitor infraslow field changes in patients with acute stroke.

METHODS

A simple motor condition was used to induce physiological cortical infraslow field changes. Five patients in a subacute state after ischemic stroke performed self-paced simple finger movements (30-second periods of finger movements, always separated by 30-second periods of rest, for a total of 15 minutes). Near-DC-magnetoencephalography signals were recorded over the contralateral primary motor cortex for the affected and unaffected hemisphere, respectively.

RESULTS

In all patients, the time courses of the contralateral cortical field amplitudes in the infraslow frequency range followed closely the motor task cycles revealing statistically significant differences between finger movement and rest periods. In 4 of 5 patients, infraslow field amplitudes were significantly stronger over the unaffected hemisphere compared with the affected hemisphere.

CONCLUSIONS

This study demonstrates that cortical infraslow activity can be recorded noninvasively in patients in the subacute state after ischemic stroke. It is suggested that near-DC-magnetoencephalography is a promising tool to also detect cortical spreading depolarization noninvasively.

Neuronal current detection with low-field magnetic resonance: simulations and methods

The noninvasive detection of neuronal currents in active brain networks [or direct neuronal imaging (DNI)] by means of nuclear magnetic resonance (NMR) remains a scientific challenge. Many different attempts using NMR scanners with magnetic fields >1 T (high-field NMR) have been made in the past years to detect phase shifts or magnitude changes in the NMR signals. However, the many physiological (i.e., the contemporarily BOLD effect, the weakness of the neuronal-induced magnetic field, etc.) and technical limitations (e.g., the spatial resolution) in observing the weak signals have led to some contradicting results. In contrast, only a few attempts have been made using low-field NMR techniques. As such, this paper was aimed at reviewing two recent developments in this front. The detection schemes discussed in this manuscript, the resonant mechanism (RM) and the DC method, are specific to NMR instrumentations with main fields below the earth magnetic field (50 microT), while some even below a few microteslas (ULF-NMR). However, the experimental validation for both techniques, with differentiating sensitivity to the various neuronal activities at specific temporal and spatial resolutions, is still in progress and requires carefully designed magnetic field sensor technology. Additional care should be taken to ensure a stringent magnetic shield from the ambient magnetic field fluctuations. In this review, we discuss the characteristics and prospect of these two methods in detecting neuronal currents, along with the technical requirements on the instrumentation.

Combined MEG and EEG methodology for non-invasive recording of infraslow activity in the human cortex

OBJECTIVE

Periinfarct depolarisation and spreading depression represent key mechanisms of neuronal injury after stroke. Changes in cortical electrical potentials and magnetic fields in the very low frequency range are relevant parameters to characterize these events, which up to now have only been recorded invasively. In this study, we proved whether a non-invasive combined MEG/EEG recording technique is able to quantitatively monitor cortical infraslow activity in humans.

METHODS

We used repetitive very slow and slow right finger movements as a physiological motor activation paradigm to induce cortical infraslow activity. Infraslow fields were recorded over the left hemisphere using a modulation-based MEG technique. EEG was performed using 16 standard Ag-Cl electrodes that covered the left motor cortex.

RESULTS

We recorded stable focal motor-related infraslow magnetic field changes in seven out of seven subjects. We also found correlating infraslow electrical potential changes in three out of seven subjects. Slow finger movements generated significantly stronger field and potential changes than very slow movements.

CONCLUSIONS

This study demonstrates the technical feasibility of combined non-invasive electrical potential and magnetic field measurements to localize and quantitatively monitor physiological, low amplitude, infraslow cortical activity in humans. This specific combination of simultaneous recording techniques allows to benefit from the specific physical advantages of each method.

SIGNIFICANCE

This combined non-invasive MEG-EEG methodology is able to provide important information on infraslow neuronal activity originating from tangentially and radially oriented sources. Moreover, this dual approach has the potential to separate neuronal from non-neuronal DC-sources, e.g., radially to the head orientated DC-currents across the skin/scalp/skull/dura occurring during cerebral hypercapnia or hypoxia.

Dynamics of cortical neurovascular coupling analyzed by simultaneous DC-magnetoencephalography and time-resolved near-infrared spectroscopy

Functional magnetic resonance imaging (fMRI) visualizes activated brain areas with a high spatial resolution. The activation signal is determined by the local change of cerebral blood oxygenation, blood volume and blood flow which serve as surrogate marker for the neuronal signal itself. Here, the complex coupling between these parameters and the electrophysiologic activity is characterized non-invasively in humans during a simple motor task using simultaneously DC-magnetoencephalography (DC-MEG), for the detection of neuronal signals, and time-resolved near-infrared spectroscopy (trNIRS), for cortical metabolic/vascular responses: over the left primary motor cortex hand area of healthy subjects DC-fields and trNIRS parameters followed closely the 30 s motor task cycles, i.e., finger movements of the right hand alternating with rest. In subjects showing a sufficient signal-to-noise ratio the analysis of variance of photon time of flight proved that the task-related trNIRS changes originated from the cortex. While onset and relaxation started simultaneously, trNIRS signals reached 50% of the maximum level 1-4 s later than the DC-MEG-signals. The non-invasive 'dual' setup helps to characterize simultaneously the two complementary aspects of the 'hemodynamic inverse problem', i.e., the coupling of neuronal and vascular/metabolic signals, in healthy subjects and provides a new analysis perspective for pathophysiological coupling concepts in diverse diseases, e.g., in stroke, hypertension and Alzheimer's disease.

DC-magnetoencephalography and time-resolved near-infrared spectroscopy combined to study neuronal and vascular brain responses

The temporal relation between vascular and neuronal responses of the brain to external stimuli is not precisely known. For a better understanding of the neuro-vascular coupling changes in cerebral blood volume and oxygenation have to be measured simultaneously with neuronal currents. With this motivation modulation dc-magnetoencephalography was combined with multi-channel time-resolved near-infrared spectroscopy to simultaneously monitor neuronal and vascular parameters on a scale of seconds. Here, the technique is described, how magnetic and optical signals can be measured simultaneously. In a simple motor activation paradigm (alternating 30 s of finger movement with 30 s of rest for 40 min) both signals were recorded non-invasively over the motor cortex of eight subjects. The off-line averaged signals from both modalities showed distinct stimulation related changes. By plotting changes in oxy- or deoxyhaemoglobin as a function of magnetic field a characteristic trajectory was created, which was similar to a hysteresis loop. A parametric analysis allowed quantitative results regarding the timing of coupling: the vascular signal increased significantly slower than the neuronal signal.

Tonic neuronal activation during simple and complex finger movements analyzed by DC-magnetoencephalography

Functional neuroimaging techniques map neuronal activation indirectly via local concomitant cortical vascular/metabolic changes. In a complementary approach, DC-magnetoencephalography measures neuronal activation dynamics directly, notably in a time range of the slow vascular/metabolic response. Here, using this technique neuronal activation dynamics and patterns for simple and complex finger movements are characterized intraindividually: in 6/6 right-handed subjects contralateral prolonged (30 s each) complex self-paced sequential finger movements revealed stronger field amplitudes over the pericentral sensorimotor cortex than simple movements. A consistent lateralization for contralateral versus ipsilateral finger movements was not found (4/6). A subsequent sensory paradigm focused on somatosensory afferences during the motor tasks and the reliability of the measuring technique. In all six subjects stable sustained neuronal activation during electrical median nerve stimulation was recorded. These neuronal quasi-tonic activation characteristics provide a new non-invasive neurophysiological measure to interpret signals mapped by functional neuroimaging techniques.

GLM-beamformer method demonstrates stationary field, alpha ERD and gamma ERS co-localisation with fMRI BOLD response in visual cortex

Recently, we introduced a new 'GLM-beamformer' technique for MEG analysis that enables accurate localisation of both phase-locked and non-phase-locked neuromagnetic effects, and their representation as statistical parametric maps (SPMs). This provides a useful framework for comparison of the full range of MEG responses with fMRI BOLD results. This paper reports a 'proof of principle' study using a simple visual paradigm (static checkerboard). The five subjects each underwent both MEG and fMRI paradigms. We demonstrate, for the first time, the presence of a sustained (DC) field in the visual cortex, and its co-localisation with the visual BOLD response. The GLM-beamformer analysis method is also used to investigate the main non-phase-locked oscillatory effects: an event-related desynchronisation (ERD) in the alpha band (8-13 Hz) and an event-related synchronisation (ERS) in the gamma band (55-70 Hz). We show, using SPMs and virtual electrode traces, the spatio-temporal covariance of these effects with the visual BOLD response. Comparisons between MEG and fMRI data sets generally focus on the relationship between the BOLD response and the transient evoked response. Here, we show that the stationary field and changes in oscillatory power are also important contributors to the BOLD response, and should be included in future studies on the relationship between neuronal activation and the haemodynamic response.

Magnetoencephalography: an investigational tool or a routine clinical technique?

Magnetoencephalography (MEG) is a relatively novel noninvasive technique, with a much shorter history than EEG, that conveys neurophysiological information complementary to that provided by EEG, with high temporal and spatial resolution. Despite its a priori, highly competitive profile, the role of MEG in the clinical setting is still controversial. We briefly review the major obstacles MEG faces in becoming a routine clinical test and the different strategies needed to bypass them. The high cost and complexity associated with MEG equipment are powerful hindrances to wide acceptance of this relatively new technique in clinical practice. The most straightforward advantage is based on the relative facility of MEG recordings in the process of source localization, which also carries some degree of uncertainty, thus partly explaining why the development of clinical applications of MEG has been so slow. Obviously, a decrease in the cost and the elaboration of semiautomatic protocols that could reduce the complexity of the studies and favor the development of consensual strategies, as well as a major effort on the part of clinicians to identify clinical issues where MEG could be decisive, would be most welcome.

Neurovascular coupling analyzed non-invasively in the human brain

Most functional brain imaging methods detect neuronal activations indirectly through the accompanying neurovascular response. Here, we demonstrate that a novel methodological approach, the combination of DC-magnetoencephalography (DC-MEG) and near-infrared spectroscopy (NIRS), allows non-invasive assessment of the dynamics of neurovascular coupling in the human brain: detecting directly slow neuronal processes (with time constants of 30s), DC-MEG revealed, even in unaveraged recordings, sustained neuronal activations at pericentral hand cortices contralateral to repetitive finger movements; these were accompanied by the ensuing local vascular response showing similar dynamical features as quantified by simultaneously recorded NIRS. This non-invasive approach opens a new avenue for the understanding of neurovascular coupling during physiological tasks as well as in diseases involving slow neuronal depolarization shifts and alterations of blood flow, such as stroke or migraine.

Controversies in neurophysiology. MEG is superior to EEG in localization of interictal epileptiform activity: Pro

Both EEG and magnetoencephalography (MEG), with a time resolution of 1 ms or less, provide unique neurophysiologic data not obtainable by other neuroimaging techniques. MEG and EEG have often been compared to each other now although the two are complementary. Now that MEG has emerged as a mature clinical technology, it is worthwhile to compare the relative strengths of each for the localization of interictal epileptiform activity and to describe the strengths of MEG relative to EEG in the localization of interictal epileptiform activity. The sources of MEG and EEG signals will first be reviewed. Issues relevant to solving the forward problem and the inverse problem in MEG and EEG will be addressed followed by a comparison of research concerning the detection and localization of interictal epileptiform activity by MEG and EEG. The emphasis will be upon techniques and software routinely used in clinical applications but some emerging areas of MEG research which are entering clinical practice will also be reviewed.

SIGNIFICANCE

MEG is a new noninvasive neurophysiologic technique which provides unique information for the clinical evaluation of patients with epilepsy, revealing aspects of neuronal function that previously could only be obtained by invasive EEG monitoring, and giving a new window for research of neuronal activity.

MEG and EEG in epilepsy

Both EEG and magnetoencephalogram (MEG), with a time resolution of 1 ms or less, provide unique neurophysiologic data not obtainable by other neuroimaging techniques. MEG has now emerged as a mature clinical technology. While both EEG and MEG can be performed with more than 100 channels, MEG recordings with 100 to 300 channels are more easily done because of the time needed to apply a large number of EEG electrodes. EEG has the advantage of the long-term video EEG recordings, which facilitates extensive temporal sampling across all periods of the sleep/wake cycle. MEG and EEG seem to complement each other for the detection of interictal epileptiform discharges, because some spikes can be recorded only on MEG but not on EEG and vice versa. Most studies indicate that MEG seems to be more sensitive for neocortical spike sources. Both EEG and MEG source localizations show excellent agreement with invasive electrical recordings, clarify the spatial relationship between the irritative zone and structural lesions, and finally, attribute epileptic activity to lobar subcompartments in temporal lobe and to a lesser extent in extratemporal epilepsies. In temporal lobe epilepsy, EEG and MEG can differentiate between patients with mesial, lateral, and diffuse seizure onsets. MEG selectively detects tangential sources. EEG measures both radial and tangential activity, although the radial components dominate the EEG signals at the scalp. Thus, while EEG provides more comprehensive information, it is more complicated to model due to considerable influences of the shape and conductivity of the volume conductor. Dipole localization techniques favor MEG due to the higher accuracy of MEG source localization compared to EEG when using the standard spherical head shape model. However, if special care is taken to address the above issues and enhance the EEG, the localization accuracy of EEG and MEG actually are comparable, although these surface EEG analytic techniques are not typically approved for clinical use in the United States. MEG dipole analysis is approved for clinical use and thus gives information that otherwise usually requires invasive intracranial EEG monitoring. There are only a few dozen whole head MEG units in operation in the world. While EEG is available in every hospital, specialized EEG laboratories capable of source localization techniques are nearly as scarce as MEG facilities. The combined use of whole-head MEG systems and multichannel EEG in conjunction with advanced source modeling techniques is an area of active development and will allow a better noninvasive characterization of the irritative zone in presurgical epilepsy evaluation. Finally, additional information on epilepsy may be gathered by either MEG or EEG analysis of data beyond the usual bandwidths used in clinical practice, namely by analysis of activity at high frequencies and near-DC activity.

A resampling approach to estimate the stability of one-dimensional or multidimensional independent components

When applying unsupervised learning techniques in biomedical data analysis, a key question is whether the estimated parameters of the studied system are reliable. In other words, can we assess the quality of the result produced by our learning technique? We propose resampling methods to tackle this question and illustrate their usefulness for blind-source separation (BSS). We demonstrate that our proposed reliability estimation can be used to discover stable one-dimensional or multidimensional independent components, to choose the appropriate BSS-model, to enhance significantly the separation performance, and, most importantly, to flag components that carry physical meaning. Application to different biomedical testbed data sets (magnetoencephalography (MEG)/electrocardiography (ECG)-recordings) underline the usefulness of our approach.

Non-invasive single-trial monitoring of human movement-related brain activation based on DC-magnetoencephalography

Neuroimaging techniques, such as fMRI, PET and near-infrared spectroscopy, monitor task-related neuronal activations in the brain indirectly through the associated neurovascular/metabolic responses. To assess the primary neuronal activations directly, magnetoencephalography was combined here with a mechanical modulation of the head-to-sensor position and signal separation via independent component analysis. In all of five subjects this approach allowed to monitor the time evolution of DC fields (<0.1 Hz) over the left hemisphere related to complex finger movements of the right hand alternating with rest periods (30 s each). Throughout the recording period of 30 min, stable task-related DC fields were recordable in a single-trial mode, i.e. without any averaging. DC-MEG opens up the possibility of analysing non-invasively cortical DC-activity also in stroke, migraine or epilepsy patients.

Near-DC magnetic fields following a periodic presentation of long-duration tonebursts

OBJECTIVES

The purpose of this study was to determine the time course of low-frequency (<0.1 Hz) magnetic field components evoked by long-duration tonebursts. The following two questions were of central interest. Does the sustained field (SF) show adaptation as described before for the sustained potential (SP)? How does the field amplitude return to the pre-stimulus baseline after stimulus offset?

METHODS

Neuromagnetic measurements were done with a 37-channel first-order gradiometer system. The stimulus was a 1 kHz toneburst of 10 s duration presented at fixed 20 s intervals. The averaged data (high-pass filtered, 0.03 Hz cut-off) were analyzed using the model of an equivalent current dipole with time-invariant location and orientation (fixed dipole).

RESULTS

In the grand average of the subjects with the best signal-to-noise ratio, the SF exhibited adaptation with a time constant of 3.6 s. After stimulus offset, the amplitude of the dipole moment dropped to a lower level within 300 ms and decayed exponentially to the baseline thereafter (time constant 2.7 s).

CONCLUSIONS

A two-component model is proposed: One component roughly follows the envelope of the stimulus, the other behaves like a leaky integrator. A better understanding of near-DC fields appears to be crucial for the understanding of the relationship between magnetoencephalography and other functional imaging techniques like functional magnetic resonance imaging and positron emission tomography.

Hyperventilation-induced human cerebral magnetic fields non-invasively monitored by multichannel 'direct current' magnetoencephalography

Self-paced hyperventilation (HV) induces slow cerebral magnetic field changes which were monitored and mapped continuously over 15 min using 49-channel DC-coupled ('direct current') magnetoencephalography (DC-MEG) based on a modulation technique. In nine/nine healthy subjects HV caused an increase (range: 1.1-6.2 pT) of the mean global DC-MEG field strength which slowly decayed after HV termination (mean time constant: 2 min). The complex HV-related field patterns were distinctly different from mainly dipolar somatosensory evoked field maps (N20m) in four/four subjects. Thus, current sources in the primary somatosensory cortex need not regularly dominate DC-field changes as had been previously considered. Rather, DC-MEG enabled the monitoring of a widely distributed HV-induced enhanced cortical excitability which may serve as model to study epileptic or post-anoxic cerebral hyperexcitability.

Independent component analysis of noninvasively recorded cortical magnetic DC-fields in humans

We apply a recently developed multivariate statistical data analysis technique--so called blind source separation (BSS) by independent component analysis--to process magnetoencephalogram recordings of near-dc fields. The extraction of near-dc fields from MEG recordings has great relevance for medical applications since slowly varying dc-phenomena have been found, e.g., in cerebral anoxia and spreading depression in animals. Comparing several BSS approaches, it turns out that an algorithm based on temporal decorrelation successfully extracted a dc-component which was induced in the auditory cortex by presentation of music. The task is challenging because of the limited amount of available data and the corruption by outliers, which makes it an interesting real-world testbed for studying the robustness of ICA methods.