Coregistered EEG/MEG for source localization in epilepsy

Abstract

The view of brain activity is determined by the method used, for example, conventional scalp electroencephalography (EEG) provides a blurred vision, subdural invasive electrodes a view like a magnifying lens, and intracerebral electrodes, a tunnel view. The development of sophisticated algorithms for source localization of magnetoencephalography (MEG) and EEG provides a new approach to recognize and localize source attributes. Intracellular and extracellular currents are the sources of electric/magnetic fields. The intracellular current (primary current) is surrounded by a magnetic field, whereas extracellular currents are due to passive volume conduction induced by the electric field. EEG and MEG are complementary recording techniques because both methods have certain advantages and disadvantages. MEG has some favorable characteristics for source localizations; for example, secondary volume currents are in general not so important and the modeling of the conductivities of the different brain tissues (gray–white matter, skull, scalp) do not play a major role. Modern MEG–multichannel systems with up to 300 channels provide a fast and contact-less measurement of the predominantly tangential components of the brain activity. On the other hand, EEG records both radial and tangential components, but secondary volume conduction plays a significant role and must be considered adequately by a realistic head model (Boundary Element Model, Finite Element Model). For source localization, special electroanatomic and physical characteristics have to be considered, because the solution of the inverse problem is not unique. Therefore, additional assumptions have to be made for the source model and the allowed source space. Very often the equivalent current dipole (ECD), which describes the center of gravity of the electrical activity, is used for the description of focal epileptic activity. This model provides in many cases a good estimation of the underlying source but a careful evaluation of the results with respect to stability and reproducibility have to be considered. In addition to the ECD, more elaborate source models, for example, multiple dipoles, beam former, or current density analysis can be used. In addition, the coherence analysis provides a better understanding of the brain dynamics. As in all functional imaging methods, the extension of activity and noise artifacts require special consideration. The depth resolution of MEG recordings depends on technical conditions such as the use of gradiometers or magnetometers. In addition to scalp EEG/MEG recordings, also simultaneously recorded electrocorticography (ECoG) and MEG recordings can be performed offering a better three-dimensional view of deeper localized epileptic activities. Source localization aims to detect focal epileptic activity within the interictal or even ictal onset of the epileptic network. In addition to the localization of epileptic activity, its spatial relationship to functional important areas (e.g., motor/speech) is a major task with regard to surgical planning. For this task a multimodal coregistration of MEG/EEG and functional magnetic resonance imaging (fMRI) as well as tractography by means of diffusion tensor imaging (DTI) can be used (Fig. 1). (A) Streamtube visualization of the right optic radiation based on diffusion tensor imaging (DTI). (B) For navigation, a three-dimensional object representing the optic radiation (wrapping the individual fibers) and two distinct magnetic source imaging (MSI) foci (red) are generated. (C) Relation of optic radiation (visualized as streamlines) to MSI foci. (D–F) Sagittal/coronal (axial view of T1-weighted images with registered DTI. Localization of focal epileptic activity is below the optic tract. (Stefan et al., 2006). The sensitivity of MEG for specific epileptic activity is approximately 70%. The MEG, compared to video-EEG monitoring and MRI, provided additional information during presurgical evaluation in 35% (Stefan et al., 2003). In patients, for whom interictal and focal epileptic seizures during long-term scalp EEG recordings were inconclusive (25 of 105 patients) MEG was able to localize in the lobe, which has been resected later on, in 44%. All these patients had an improvement of seizure frequency postoperatively (six patients were seizure-free; five patients had a reduction >50%) (Paulini et al., 2007). When comparing noninvasive magnetic source imaging and invasive seizure-onset localization, a positive predictive value of magnetic source imaging (MSI) for seizure localization was 82–90% (Knowlton et al., 2006). MEG predicted outcome following surgery for intractable epilepsy in children with normal or nonfocal MRI findings (RamachandranNair et al., 2007). There are special situations during presurgical evaluation in which MSI is helpful: for example, patients after surgery, multiple stage surgery, patients with missing cooperation for prolonged invasive recordings or contraindications for invasive recordings. Furthermore, presurgical invasive recordings can be guided by MEG and in some cases might be even omitted. The combined multimodal use of MEG/EEG, MRI, fMRI, single photon emission computed tomography (SPECT), and functional neuronavigation can be used for functional preservation in epilepsy surgery (Ganslandt et al., 1999). The author has no conflict of interest to disclose.