Abstracts

Rationale: Magnetoencephalography (MEG) has significant advantages for localization due to its higher temporal and spatial resolution along with insensitivity to tissue inhomogeneities (e.g. skull, scalp), compared with electroencephalography (EEG). Localization accuracy is highly dependent on the signal to noise ratio (SNR) which, despite magnetically shielded rooms, is difficult to control. In most mapping and cognitive-related protocols, averaging is employed to improve SNR. During spontaneous MEG recordings in epilepsy patients, averaging of interictal activity is not desirable, and alternative methods of noise reduction are sought. Since inception of our MEG laboratory we have post-processed our MEG data in order to reduce magnetic noise to a level permitting adequate source localization with equivalent current dipole (ECD) methods. Methods: In consecutive intractable epilepsy patients (approximately one per week over three months) EEG and MEG were simultaneously recorded for 30 - 40 minutes. Spontaneous MEG data were acquired with a 204 planar gradiometer MEG system (Neuromag, Helsinki, Finland), while a minimum of 21 scalp EEG electrodes were placed according to the international 10-20 system. Data was reviewed to identify interictal spikes. The data was bandpass-filtered between 0.01 and 330 Hz and sampled at 1000 Hz. All of the data was also reviewed after processing with a spatiotemporal signal space separation (tSSS) method. Number and goodness of fit (GOF) of calculated source dipoles was compared before and after processing with tSSS. Results: In our pilot study, the number of dipoles after tSSS was generally larger than that before tSSS and the GOF after tSSS was higher than that before tSSS. In one patient with a vagal nerve stimulator (VNS), there were no acceptable dipole solutions before tSSS filtering but sufficient acceptable dipoles after tSSS filtering. Conclusions: Most of the published work in MEG has been carried out on volunteer research subjects in (non-hospital) magnetically quiet environments. In contrast, clinical patients with epilepsy cannot be expected to cooperate as fully, and their MEG recordings are likely to include more interference. Processing the MEG recordings with a method to eliminate artifact arising from outside the brain, such as tSSS, can significantly improve the recordings. In some cases, this improvement can mean the difference between satisfactory dipole fits vs no localization possible.