John S. Ebersole

A standardized boundary element method volume conductor model.

OBJECTIVES We used a 3-compartment boundary element method (BEM) model from an averaged magnetic resonance image (MRI) data set (Montreal Neurological Institute) in order to provide simple access to realistically shaped volume conductor models for source reconstruction, as compared to individually derived models. The electrode positions were transformed into the models coordinate system, and the best fit dipole results were transformed back to the original coordinate system. The localization accuracy of the new approach was tested in a comparison with simulated data and with individual BEM models of epileptic spike data from several patients. METHODS The standard BEM model consisted of a total of 4770 nodes, which describe the smoothed cortical envelope, the outside of the skull, and the outside of the skin. The electrode positions were transformed to the model coordinate system by using 3-5 fiducials (nasion, left and right preauricular points, vertex, and inion). The transformation consisted of an averaged scaling factor and a rigid transformation (translation and rotation). The potential values at the transformed electrode positions were calculated by linear interpolation from the stored transfer matrix of the outer BEM compartment triangle net. After source reconstruction the best fit dipole results were transformed back into the original coordinate system by applying the inverse of the first transformation matrix. RESULTS Test-dipoles at random locations and with random orientations inside of a highly refined reference BEM model were used to simulate noise-free data. Source reconstruction results using a spherical and the standardized BEM volume conductor model were compared to the known dipole positions. Spherical head models resulted in mislocation errors at the base of the brain. The standardized BEM model was applied to averaged and unaveraged epileptic spike data from 7 patients. Source reconstruction results were compared to those achieved by 3 spherical shell models and individual BEM models derived from the individual MRI data sets. Similar errors to that evident with simulations were noted with spherical head models. Standardized and individualized BEM models were comparable. CONCLUSIONS This new approach to head modeling performed significantly better than a simple spherical shell approximation, especially in basal brain areas, including the temporal lobe. By using a standardized head for the BEM setup, it offered an easier and faster access to realistically shaped volume conductor models as compared to deriving specific models from individual 3-dimensional MRI data.

Intracranial EEG substrates of scalp EEG interictal spikes.

Summary:  Purpose: To determine the area of cortical generators of scalp EEG interictal spikes, such as those in the temporal lobe epilepsy.

Localization of Temporal Lobe Foci by Ictal EEG Patterns

Identifying patients whose complex partial seizures originate in temporal neocortex rather than in hippocampus is important because such patients have less favorable outcomes with standard anteromesial temporal resections. We reviewed scalp‐recorded ictal EEGs of 93 epilepsy surgery candidates who either underwent intracranial EEG monitoring (n= 58) or who were referred directly for temporal lobectomy (n= 35). We defined seven patterns of early seizure discharges, grouped patients according to their seizure pattern, and correlated these with the site of seizure onset determined by intracranial EEG. Categorization by seizure pattern was also compared with brain magnetic resonance imaging (MRI) findings and intracarotid amobarbital (Wada) testing. An initial, regular 5‐ to 9‐Hz inferotemporal rhythm (type 1A) was most specific for hippocampal‐onset seizures. Less commonly, a similar vertex/parasagittal positive rhythm (type 1B) or a combination of types 1B and 1A rhythms (type 1C) was recorded. Seizures originating in temporal neocortex were most often associated with irregular, polymorphic, 2‐ to 5‐Hz lateralized activity (type 2A). This pattern was commonly followed by a type 1A theta rhythm (type 2B) or was preceded by repetitive, sometimes periodic, sharp waves (type 2C). Seizures without a clear lateralized EEG discharge (type 3) were most commonly of temporal neocortical origin. These associations between type of seizure pattern and probable site of cerebral origin were statistically significant. MRI and Wada testing did not have as much specificity as ictal patterns in differentiating among seizure origins. We conclude that the initial pattern of ictal discharge on scalp EEG can assist in distinguishing seizures of temporal neocortical onset from those of hippocampal onset. This information can be used to identify patients for invasive monitoring.

Intracranial EEG substrates of scalp ictal patterns from temporal lobe foci

Summary: Purpose: To determine the intracranial EEG features responsible for producing the various ictal scalp rhythms, which we previously identified in a new EEG classification for temporal lobe seizures.

DEFINING EPILEPTOGENIC FOCI : PAST, PRESENT, FUTURE

There is a direct relationship between the geometry (location, area, and orientation) of cortex-generating epileptiform discharges and resultant spike or seizure voltage fields at the scalp. Epileptogenic foci have been localized traditionally with EEG by identifying the negative field maximum (e.g., a phase reversal between adjacent bipolar channels). However, it is the shape of the entire voltage field over the head, including both negative and positive maxima, which provides information necessary to characterize the focus properly. Source location and orientation can be inferred from spike or seizure voltage topography, however, three-dimensional visualization can be obtained from mathematical source models, such as an equivalent dipole. Recent investigations have shown that dipole models can identify the location of epileptogenic foci with sub-lobar precision. Accuracy is enhanced by using additional electrodes, particularly on the lower half of the head, and by measuring their location. Realistic head models obtained from three-dimensional reconstructions of MR images can overcome errors introduced by simple spherical models of the cranium. Co-registering EEG voltage topography and source models with a patients own cerebral anatomy will make EEG an unparalleled functional imaging technique for defining epileptogenic foci.

An evaluation of ambulatory, cassette EEG monitoring: I. Montage design

Ambulatory EEGs using cassette tape recorders have only four channels. Questions have been raised about the limitations this imposes on detection of focal interictal epileptiform events. A review of EEGs from inpatients undergoing prolonged monitoring demonstrated an overwhelming representation of these abnormalities in the frontal and anterior temporal scalp regions. Three-channel montages were designed to sample these regions and were tested by simultaneously recording them with multichannel montages. Successful montages combining a frontal transverse and two longitudinal channels adequately detected 74 to 100% of interictal epileptic events.

EEG dipole modeling in complex partial epilepsy.

SummaryVisual inspection and qualitative impressions of clinical EEG abnormalities are being replaced by quantitative characterization of scalp voltage fields and dipole modeling of underlying cerebral sources. Three approaches have been used in the analysis of focal spikes of complex partial epilepsy. 1) Instantaneous, single dipole, inverse solutions for the voltage topography of the spike peak have revealed two distinct equivalent dipole configurations in the brain lobe beneath the negative extreme - radial and oblique (mixed radial and tangential). Only radial dipoles have been found for frontal and fronto-central spikes, while either type have been found for temporal and occipital spike foci. 2) Dipole stability can be assessed by an inspection of sequential instantaneous solutions encompassing the spike complex or by calculating the standard deviation of dipole location (x,y,z) and orientation (elevation, azimuth) parameters during this period. Two-thirds of spike dipoles of the radial type and essentially all of the oblique equivalent dipoles were found to be stable, whereas one-third of the radial dipoles were unstable in position or orientation. 3) Spatio-temporal analysis can identify multiple underlying sources and their potentials. Modeling separate radial and tangential dipoles over the course of the spike has revealed a composite character for spike fields with oblique dipoles and often has defined leads or lags in activity that suggested propagation between infero-mesial and lateral temporal cortex. Correlations with clinical and intracranial EEG data suggest that patients with mesial temporal sclerosis have spikes with oblique and stable equivalent dipoles; patients with discrete cortical lesions have spikes with radial and stable dipoles; patients with extensive or multi-focal cortical insults have spikes with radial and unstable dipoles.

Models of brain sources

SummaryTwo categories of models are available for the functional imaging of scalp recorded electric brain activity: single-time-point and spatio-temporal. Instantaneous models require strict assumptions that do not conform with the underlying physiology, because they rely on the few voltage differences measured at only one sampling point. Spatio-temporal models create a spatial image of discrete multiple sources and a temporal image of source current wave forms which reflect the time course of the local activity in circumscribed brain areas at a macroscopic level. The spatial image may be limited in accuracy because it depends both on model and data, but it can be validated by scanning the brain with regional dipole sources. In many cases of temporal lobe epilepsy, for example, interictal spikes can be described adequately by as few as two equivalent dipoles, which image the vertical source current arising from the medio-basal aspect of the temporal lobe and the horizontal source current from its lateral surface.

Magnetoencephalography/Magnetic Source Imaging in the Assessment of Patients with Epilepsy

Summary: Magnetoencephalographic (MEG) dipole source localization is a particularly promising new tool for noninvasive presurgical evaluation of epileptogenic foci. It is potentially more accurate than EEG localization techniques because magnetic fields are not attenuated or distorted by the skull and scalp, which allows cerebral sources to be modeled more simply. MEG spike and seizure sources are routinely co‐registered with the patients brain MRI for clinical interpretation. This has been called magnetic source imaging. Numerous studies have shown that MEG localization of foci agreed with lesion position, depth electrode and ECoG data, PET and MRI findings, and surgical success. The recent development of whole head sensor arrays has greatly enhanced the ease with which epileptiform magnetic fields can be recorded and analyzed.

Noninvasive localization of epileptogenic foci by EEG source modeling

Localization of the epileptogenic focus is a ratelimiting factor in evaluation of patients with medically uncontrolled partial seizures for epilepsy surgery. Considerable delay, expense, and possible morbidity are incurred if intracranial EEG monitoring becomes necessary. Accordingly, improving noninvasive presurgical localization has become an area of increased interest. Anatomic imaging now plays a significant role in this regard. However, an increasing number of patients are being considered who do not have well-defined MRI lesions or atrophic regions. Functional imaging techniques, such as positron emission tomography (PET), single-photon emission computed tomography (SPECT), and functional MRI (fMRI) are also being used to help identify the epileptogenic focus. Unfortunately, these methods image secondary phenomena rather than measuring epileptic brain activity directly. Furthermore, none of these functional techniques has sufficient temporal resolution to distinguish the origin of spike or seizure activity as opposed to that which results from propagation. We therefore continue to rely on the analysis of interictal spikes and ictal rhythms recorded by scalp EEG for noninvasive functional localization. Unfortunately, traditional EEG analysis by visual inspection is simplistic at best and misleading at worst. Characterization of an epileptogenic focus may be limited to identifying the electrode recording the maximal negative potential. Such localization is based on the inaccurate assumption that the cortical generator must underlie the scalp EEG field maximum. If our use of EEG is to advance, we must progress beyond simple description of waveforms and pursue the activity and location of underlying cerebral sources. Computer-assisted EEG analysis greatly aids this effort. Voltage topography of spike and seizure potentials is the basis for epileptogenic focus localization with biophysical source models, such as the equivalent dipole. These models are more correctly based on the assumption that the relative location of both voltage maxima (negative and positive) and the contours of the voltage fields between them convey information concerning source location, orientation, and propagation. I argue here that spikeheizure foci are modeled sufficiently well by equivalent EEG dipoles to provide significantly improved localization over traditional EEG interpretation, that interpretation of EEG dipoles can be enhanced by co-registering them with 2D and 3D MRI of the patient’s brain, and that dipole location accuracy is improved, particularly for basal foci, by employing realistic head models. By using EEG source modeling, I believe that the number of patients requiring chronic intracranial EEG monitoring will be reduced and that the accuracy of electrode placement will be improved in patients who require invasive recording.