Abstracts

Rationale: Magnetoencephalography (MEG) is a well accepted non-invasive pre-surgical diagnostic evaluation in patients with intractable focal epilepsy. A commonly used clinical MEG instrument contains 204 planar gradiometers and 102 magnetometers. A gradiometer is a two-loop coil designed to eliminate distant noise from the environment while enhancing nearby signals, e.g. from epileptic generators. A magnetometer is a single loop coil designed to theoretically detect signals from the deeper sources, but at the expense of higher susceptibility to environmental noise and the greater need of stronger magnetic shielding. Although the MEG instrument collects data on both sensor types concurrently, in practice many institutions use only the gradiometers for the detection and localization of epileptic sources. The purpose of this study was to determine the clinical utility of magnetometers by comparing their ability to localize sources in patients with focal epilepsy when combined with gradiometers.Methods: Measurements from six patients, performed inside a magnetically shielded room using a helmet-shaped whole-head MEG system (Elekta Neuromag, Finland), were analyzed; EEG (10-20 system) data were also simultaneously recorded. All MEG epileptiform discharges from a total of approximately six hours of recording were visually identified, then processed using the vendor s built-in software, to generate equivalent current dipoles (ECDs). For each ECD, baseline noise statistics were calculated from background data occurring in a nearby time epoch, and the empty room noise was projected out. The following parameters were recorded: Cartesian location, goodness of fit (GOF), reduced chi-square, and confidence volume. For each dipole source localization, identical sensor regions were employed, i.e. all 153 sensors (ALL) on either the left or right side of the helmet, or just the corresponding 102 gradiometers (GRAD).Results: From the six patients combined, 101 MEG epileptiform discharges were analyzed. ECDs were localized to mesial occipital, mesial frontal, mesial parietal, superior temporal sulcus, precentral and postcentral sulci, frontal operculum, parietal operculum and vertex. The reduced chi-squares were 1.33 ( 1.05) for ALL, and 1.42 ( 1.19) for GRAD, with corresponding GOFs of 79% ( 8.5), and 78%( 8.7). The average change in source location was 1.86mm ( 1.51), and the confidence volume ratio was 1.74 ( 0.20), compared to a theoretical ratio of 1.84. Conclusions: The addition of magnetometers does not appear to substantially improve the source localization of the gradiometer-only solutions. Further investigation is needed underway to identify the regions of the brain where magnetometers may show to be superior to gradiometers, especially in the deep regions of the brain.