Abstracts

Rationale: Magenetoencephalography (MEG) can be easily corrupted by interfering sources of magnetic fields. The vagal nerve stimulator (VNS) in epilepsy patients, produces very strong magnetic artifacts. VNS devices are frequently encountered in patients with epilepsy that is medically refractory epilepsy. In many centers, patients with VNS are excluded from MEG due to magnetic noise. The recent development of a temporally-extended signal space separation algorithm (tSSS) has made it possible to routinely record from these patients. We have previously reported the benefit of tSSS processing of spontaneous MEG recording (Jin et al, Epilepsia 2008;49 S7; p199). In this study, we set out to ascertain the benefit of the tSSS algorithm in the recording and localization of (averaged) evoked magnetic field data.Methods: All patients referred to the MEG laboratory in the Cleveland Clinic from its inception in 2008 until 2011 were reviewed. Of 280 patients, 24 had VNS, in whom median nerve somatosensory evoked fields (SEF) were recorded from 42 sides. SEF responses were measured using a 306-channel whole-head MEG system (Elekta Ltd, Helsinki, Finland). For both the no-tSSS and the tSSS data, a rejection threshold of approximately 3000 fT was applied to the individual SEF responses, then approximately 200 trials were averaged. Prominent peaks were visually identified approximately 20 ms after stimulus. The noise covariance matrix was obtained from the time period starting 3-4 ms after the stimulus (i.e. when the stimulus artifact had disappeared) and ranging to 10 ms before the upslope of the N20m, and equivalent current dipole (ECD) sources were estimated at the peak. We classified SEF dipoles as reliable if: A) the location of the dipole was in the expected location in the central sulcus contralateral to the side of stimulation, and B) the goodness of fit value (GOF) was greater than 80%. We analyzed the results of using or not using tSSS in two ways: 1) We counted the number of responses (sides) which produced reliable dipoles in each dataset (i.e. no-tSSS and tSSS data). 2) For those datasets where, from the same stimulation, both no-tSSS and tSSS data produced reliable dipoles, we compared the GOFs and the 95% confidence volumes (CV) (mm3). Statistical differences in the GOF and CV between with/without tSSS were determined by Wilcoxon signed-rank test. Results: Only 13 (31%) responses had reliable dipoles in the no-tSSS group, while all 42 (100%) had reliable dipoles in the tSSS group. Of the 13 responses which were reliable with or without tSSS processing, their dipoles had significantly higher GOF and lower CV after tSSS processing than without it (p values less than 0.05).Conclusions: The tSSS algorithm quantitatively improves dipole fitting of known sources in VNS patients, such as the cortical generators of the median nerve sensory response. This algorithm permits satisfactory MEG testing in the relatively commonly encountered epilepsy patient with an implanted VNS.