Multi-Resolution FOCUSS: a source imaging technique applied to MEG data.

A variety of techniques are available for imaging magnetoencephalographic (MEG) data to the corresponding cortical structures. Each performs a functional optimization that includes mathematical and physical restrictions on source activity. Unlike other imaging techniques, MR-FOCUSS (Multi-Resolution FOCal Underdetermined System Solution) utilizes a wavelet statistical operator that allows spatial resolution to be chosen appropriately for focal or extended sources. Control of focal imaging properties is achieved by specifying P in an l(P) norm distribution template used to construct the wavelets. In addition, incorporation of a multi-resolution wavelet operator desensitizes the mathematical algorithm to noise, (regularization). Like the FOCUSS imaging technique, an initial estimate of cortical activity is recursively enhanced to obtain the final high resolution imaging results. Studies of model MEG data representing all regions of a realistic cortical model are performed to quantify MR-FOCUSS imaging properties. These modeled data studies included single and multiple dipole sources as well as an extended source model. Thus, MR-FOCUSS is found to be very effective for imaging language processing for pre-surgical planning and provides a high-resolution method to image sequential activation of multiple correlated sources involved in language processing.

MEG localization of language-specific cortex utilizing MR-FOCUSS.

OBJECTIVE: To demonstrate noninvasive localization of cognitive cortical areas involved in language processing with magnetoencephalography (MEG) interpreted by multiresolution FOCUSS (MR-FOCUSS), a current density imaging technique. METHOD: MEG data were collected during verb-generation and picture-naming tasks from 18 right-handed control subjects and 24 right-handed patients with epilepsy. RESULTS: The averaged epic data from the verb-generation task, analyzed by MR-FOCUSS, showed initial activation in the left supramarginal gyrus, superior temporal gyrus, and angular gyrus at 239 +/- 31 ms in all subjects, consistent with other language mapping studies. Average amplitude of underlying cortical sources was approximately 452 pAm. The averaged epic data from the picture-naming task, analyzed by MR-FOCUSS, showed activation in the left inferior frontal gyrus (IFG) area starting at 436 +/- 40 ms in all subjects. Average amplitudes of underlying cortical sources were approximately 380 pAm. CONCLUSION: The time course of neuronal language processing can be imaged noninvasively with millisecond resolution by magnetoencephalography using the multiresolution FOCUSS technique.

ICA methods for MEG imaging.

Activity of individual cortical sources cannot be uniquely imaged when MEG data is a sequence of complex spatial patterns of multiple cortical sources. Auxiliary constraints integrated into the imaging equations are required to remove the mathematical ambiguity. Therefore, it is important to adapt source separation techniques to MEG imaging. It is much easier to accurately image field patterns of isolated brain electric sources. We demonstrate how a combination of second and fourth order Independent Component Analysis (ICA) methods can be used to remove noise and isolate source activity for improved MEG imaging accuracy. A second order ICA technique was used to extract respiratory and eye movement artifact by exploiting cross-correlation differences over time between cortical sources and artifact. For brain electric source separation, a fourth order ICA technique that quantified probabilities of simultaneous source activity was used to separate brain electric sources characterized by bursts of oscillatory circuit activity.

Magnetoencephalographic fields from patients with spontaneous and induced migraine aura.

We investigate and characterize the magnetoencephalographic waveforms from patients during spontaneous and visually induced migraine aura. Direct current neuromagnetic fields were measured during spontaneous onset of migraine auras in 4 migraine patients, and compared with recordings from 8 migraine-with-aura patients and 6 normal controls during visual stimulation of the occipital cortex. Complex direct current magnetoencephalographic shifts, similar in waveform, were observed in spontaneous and visually induced migraine patients, but not in controls. Two-dimensional inverse imaging showed multiple cortical areas activated in spontaneous and visually induced migraine aura patients. In normal subjects, activation was only observed in the primary visual cortex. Results support a spreading, depression-like neuroelectric event occurring during migraine aura that can arise spontaneously or be visually triggered in widespread regions of hyperexcitable occipital cortex.

Two dimensional inverse imaging (2DII) of current sources in magnetoencephalography.

A new magnetoencephalographic (MEG) technique for imaging the cortical distribution of neuronal activity is described. An iterative algorithm is employed, which successively alters an initial estimate of cortical source structure until it corresponds to the measured magnetic field data. In this new technique, the continuum of electrical activity across the cortical surface is modeled as a dense grid of thousands of single equivalent current dipoles. MEG imaging of both compact and extended sources is facilitated by a wavelet-like transformation of the source space into a sequence of successively smaller composite source structures. Two of these composite source structures are combined during each iterative step to generate an improved estimate of the cortical source structure. Thus, inversion of the complete gain matrix corresponding to thousands of cortical sources is not performed. The technique requires only moderate PC based resources even for very large source grids. In contrast to minimum norm MEG imaging methods, this new algorithm is insensitive to random noise in the data. If available, prior knowledge of source structure from other imaging techniques, such as PET, MRI and fMRI, is easily incorporated as additional constraints on the source structure solution. Source images solutions corresponding to simulated data are presented. In addition, the technique is applied to source imaging of real MEG data incorporating cortical structure from volumetric MRI data. These results demonstrate the capability of our new technique for imaging combinations of compact and extended source structures.

Analysis of MEG signals of spreading cortical depression with propagation constrained to a rectangular cortical strip. I. Lissencephalic rabbit model.

Magnetic fields arising from the rabbit cortex during spreading cortical depression (SCD) were measured in order to study the currents in the neocortex during SCD. SCD was constrained to propagate in a rectangular cortical strip perpendicular to the midline. This simplified in vivo cortical preparation enabled us to correlate magnetoencephalographic (MEG) signals to their underlying currents within the cortical strip. The propagation of SCD was monitored with an array of electrodes placed along the strip. The propagation speed for SCD in the lissencephalic rabbit brain was 3. 5+/-0.3 mm/min (mean+/-S.E.M., n=14). Slow, quasi-dc, MEG signals were observed as the SCD entered into the longitudinal fissure. The currents giving rise to the MEG signals were perpendicular to the cortical surface and directed from the surface to deeper layers of the cortex. A distributed dipolar source model was used to relate the data to the underlying cortical current. The moment of the single equivalent current dipole source was 38+/-9 nA-m (n=17). This study clarified the nature of the cortical currents during SCD in a lissencephalic in vivo preparation.

Analysis of MEG signals of spreading cortical depression with propagation constrained to a rectangular cortical strip. II. Gyrencephalic swine model.

Currents produced during spreading cortical depression (SCD) in a gyrencephalic species (swine) were studied with magnetoencephalography (MEG) and electrocorticography (ECoG). SCD, initiated using electrical stimulation of the cortex, was constrained to propagate within a rectangular cortical strip in order to simplify the interpretation of the underlying currents. The ECoG signals monitored along the strip revealed that SCD propagated from an initiation site on the gyrus at a rate of 7.9+/-3.2 mm/min (n=23), entered the deep coronal sulcus and in most cases emerged from the other side of the sulcus, continuing to propagate across the next gyrus at a rate of 5.9+/-2.7 mm/min (n=22). The apparent propagation velocity within the sulcus was reduced to 1.7+/-0.8 mm/min (n=21). Strong MEG signals were observed as SCD entered the sulcus. The direction of magnetic field was opposite for SCD's on opposite banks of the sulcus. The currents were directed from a superficial layer to deeper layers of the cortex. The characteristics of SCD and associated MEG patterns from a gyrencephalic species may be similar to those in human patients during migraine aura.

Modeling of spreading cortical depression using a realistic head model.

Barkley and colleagues in 1990 reported large amplitude waves (LAWs) in time series magnetoencephalography (MEG) recordings from migraine patients and inferred that these LAWs arose from spreading cortical depression (SCD). SCD propagates slowly across the cortex in all species in which it has been observed. Previously, we reported that LAWs could be simulated and compared with the recorded signals using the four-sphere model (Wijesinghe and Tepley 1997). We showed that LAWs could arise from the propagation of SCD across a sulcus. In this paper, we model LAWs using a realistically shaped head model based on magnetic resonance images (MRI) (Roth et al. 1993). Simulated signals using this model are similar to the recorded signals. In this model, current dipoles represent the excitable neurons in the cortex and magnetic fields created by these individual dipoles are calculated. The magnetic field arising from the excited area of cortex is obtained by summing the fields due to these individual dipoles.

A four sphere model for calculating the magnetic field associated with spreading cortical depression.

In our previous model, we ascertained that the large amplitude waves (LAWs), reported by Barkley and coworkers (1990) in time series magnetoencephalography (MEG) recordings from migraine patients, could be simulated and compared with the recorded signals using a simple plane volume conductor model (Tepley and Wijesinghe 1996). In this paper, we model LAWs using the help of more complicated yet reliable four-sphere model. This mathematical model furthermore assumes that the LAWs arise from propagation of Spreading Cortical Depression (SCD) across a sulcus and these simulated signals are more similar to the recorded signals than the ones we obtained from our previous model. SCD propagates slowly across the cortex in all species in which it has been observed. In our model, current dipoles represent the excitable neurons in the cortex and magnetic fields created by these individual dipoles are calculated using a four-sphere model. The magnetic field arising from the excited area of cortex is obtained by summing the fields due to these individual dipoles. Sulci shapes are represented by simple mathematical formulae.

A technique for sequential measurements of DC neuromagnetic fields.

Magnetoencephalography (MEG) has the capability of detecting DC neuromagnetic fields which might arise from migraine, stroke, head trauma, etc. However, such fields are not readily measured in serial fashion for technical reasons, including arbitrary neuromagnetometer baselines, residual DC fields which vary from place to place inside conventional magnetically shielded rooms, and especially the high sensitivity of such measurements to small changes in distance between the probe and head. A technique is described to carry out such measurements by raising and lowering the subject under the probe and calibrating the DC shift against the amplitude of spontaneous activity.

A dipole model for spreading cortical depression.

Spreading Cortical Depression (SCD) is the hyper-excitation, followed by extreme suppression of spontaneous electrical activity in the cortex. This work models SCD propagation using current dipoles to represent excitable pyramidal cells. An area of cortex, either gyrus or sulcus, supporting SCD is represented by surface dipoles oriented perpendicular to the surface. Magnetic fields created by these individual surface dipoles are calculated using the Biot-Savart law. We have assumed a plane volume conductor to represent the sulcus to simplify the mathematical derivation. The sources included in cortical surface area of 10(-4)mm2 is represented by a signal dipole. The magnetic field arising from the entire excited area of the cortex is obtained by summing the fields due to these individual dipoles. The simulated waveforms suggest that the shapes, amplitudes, and durations of the SCD signals depend on the size of the active area of cortex involved in SCD, as well as the location and orientation of the detector. Using this dipole model, we are able to simulate the Large Amplitude Waves (LAWs) similar to those observed by Barkley et al. (1990) while measuring spontaneous activity from migraine headache patients using the assumption that these LAWs arise from propagation of SCD across a sulcus. The shape of the simulated LAW waveform is strongly influenced by the relationships between the detector location and orientation, the propagation direction of the SCD wave, and the orientation of the sulcus.

Central neurogenic mechanisms of migraine.

Evidence obtained from clinical studies, magnetoencephalography, and 31P magnetic resonance spectroscopy indicates that spreading depression is the underlying basis of migraine aura. Magnetoencephalographic and 31P magnetic resonance spectroscopic evidence also exists to explain interictal central neuronal hyperexcitability in migraine sufferers. A low intracellular brain magnesium concentration may be the link between the physiologic threshold for migraine and the attack itself.

Reexamination of gender differences in the source location of N1m.

Gender differences in the source location of the auditory evoked field (AEF) component N1m have been reported previously in a small group of subjects. The present study was conducted to evaluate further the existence of gender differences in a larger sample. Neuromagnetic recordings of AEFs were obtained from young normal hearing subjects using a 1000-Hz tone burst presented at 60 dB hearing level (HL). No significant gender-related differences were observed for the N1m peak latencies following left and right ear stimulation. The stimulating parameters and larger sample used in the present study size may account for the difference in gender effects observed between the two studies.

Evidence for multiple generators in evoked responses using finite difference field mapping: auditory evoked fields.

Electric potential maps and magnetic field maps have been used to study brain electrical activity. During the temporal course of an evoked cortical response, the electrical activity of specific neuronal subpopulations change in a sequential manner giving rise to measurable electrical potentials and magnetic fields. For these potentials and fields, both the amplitude and rate of amplitude change have characteristic, time-dependent waveforms. Presently, amplitude waveforms from multiple locations are used to generate magnetic field and electric potential maps which have been found to be useful in understanding the activity of the neurons which give rise to these maps (Romani 1990). This paper introduces a data transformation technique which results in a derived map that we have termed a "finite difference field map" (FDFM). This mapping technique provides information associated with the rate at which the amplitude of the neuronal electric activity changes. In this paper, some advantage of FDFM analysis are illustrated by application of this technique to the study of the auditory evoked cortical field (AECF) N1m waveform. Using data obtained from normal subjects it will be demonstrated that application of the FDFM technique allows the localization of the primary N1m source at an earlier latency than is possible using the conventional waveform data. The source location determined at an early latency by FDFM analysis was identical to that obtained at later from the conventional field data. These data suggest that the primary N1m source is stationary. In addition, analysis of the time sequence of FDFM field maps contains evidence of a second spatially separate source which is co-active with primary N1m source.

Occurrence of auditory evoked field (AEF) N1M and P2M components in a sample of normal subjects.

A great deal of information about the characteristics of components N1m and P2m of the auditory evoked cortical fields (AEFs) has accumulated since the late 1970s. However, a number of fundamental issues have not been addressed. For instance, some previous investigators have suggested that P2m is present consistently in normals, whereas others suggest that when present, P2m is small in amplitude. Preliminary observations in this laboratory suggested that P2m is not observed consistently in normal subjects. Therefore, the specific aim of this investigation was to estimate the frequency of occurrence of the contralaterally recorded N1m and P2m after stimulation of right and left ears in normal subjects. Results demonstrated N1m to be present in all circumstances. Consistent with our preliminary observations, P2m was often absent, and moreso after stimulation of the left ear.

Magnetoencephalography of focal cerebral ischemia in rats.

BACKGROUND AND PURPOSE: The purpose of this study was to use magnetoencephalography to record magnetic field changes in the brain during middle cerebral artery occlusion. METHODS: A direct-current electrocorticogram (two channels) and a direct-current magnetoencephalogram (seven channels) were simultaneously recorded from five rats subjected to middle cerebral artery occlusion for 1-2 hours. RESULTS: Direct-current electrocorticographic and direct-current magnetoencephalographic signal deflections were observed after the onset of middle cerebral artery occlusion and occurred repeatedly throughout the ischemic period, with a mean +/- SD time interval of 12 +/- 5 minutes. A one-to-one correspondence of the electrocorticographic and magnetoencephalographic signal deflections was demonstrated. CONCLUSIONS: Direct-current magnetoencephalography can provide a new noninvasive technique for studying depolarization and/or spreading depression in focal cerebral ischemia.

Auditory evoked cortical magnetic field (M100-M200) measurements in tinnitus and normal groups.

Recently, Hoke et al. (1989) and Pantev et al. (1989) demonstrated that the auditory evoked cortical magnetic field (AECMF) M100 component was larger, and M200 was smaller and occurred later in subjects with unilateral tinnitus compared with normal subjects. These group amplitude differences resulted in an M200/M100 amplitude ratio that was smaller for the subjects with tinnitus. The purposes of the present investigation were to: 1) extend the observations of Hoke et al. (1989), and, 2) determine whether contralateral AECMF differences existed following stimulation of the non-tinnitus and tinnitus ears of patients with tinnitus. Neuromagnetic AECMF recordings were recorded from 25 young normal hearing and 14 patients with unilateral tinnitus and hearing loss. The results failed to support the findings of Hoke et al. (1989). Specifically, there is no evidence suggesting that the M100 amplitude is larger, the M200 latency later, or, the M200/M100 amplitude ratios smaller, when the two samples are compared. Additionally, there were no differences in the amplitudes or latencies of M100 or M200 when results from stimulation of the tinnitus and non-tinnitus ears of tinnitus subjects were compared.

Magnetic fields associated with anoxic depolarization in anesthetized rats.

We have performed simultaneous measurements of the DC-magnetoencephalogram (DC-MEG) and DC-electrocorticogram (DC-ECoG) in rats (n = 6) subjected to 90 s of reversible anoxia. The onset of major shifts of electric and magnetic signals occurred at 52 +/- 18 (S.D.) and 68 +/- 14 (S.D.), respectively, and reached a peak at 83 +/- 27 and 102 +/- 19 (S.D.) s, respectively, after termination of mechanical ventilation. DC-ECoG signal deflections were always associated with DC-MEG deflections. The time of onset and peak signals in both DC-MEG and DC-ECoG changes caused by asphyxia were highly correlated (r + 0.83, 0.94; P less than 0.05, 0.001; respectively). Our observations suggest that the non-invasive technique of DC-MEG is reliable and may provide insight into the mechanisms of anoxic cerebral depolarization.