Abstracts

Rationale: Literature on the yield of ictal magnetoencephalography (MEG) is limited. Use of the conventional single equivalent current dipole analysis (sECD) has been proposed in some of the published case series. However, sECD poses a conceptual challenge for certain types of ictal onset, specifically those characterized by paroxysmal fast activity. In this study we compare the standard sECD method to a tailored rhythm-oriented method that takes into account the fast activity of interest at onset and disregards other rhythms. Thus we propose the use of dynamic statistical parametric mapping dSPM estimation of sources for a narrow (seizure-specific) frequency band of interest. Methods: 377 consecutive MEG studies performed at Cleveland Clinic between 2008 and 2011 were screened and patients who had seizures during MEG recordings were identified. Sixteen patients had seizures with onsets characterized by paroxysmal fast activity. Of these patients, 12 had subsequent epilepsy surgery with postoperative follow-up of >6 months. Dipoles by sECD were accepted when standard modeling parameters were met. Our proposed method of analysis is based on identification of the ictal rhythm in a time frequency space using Morlet wavelet transformation. We then identify the dominant frequency at onset in terms of signal power. We then verify the stability in the frequency domain and evolution in time domain by mapping the band of interest onto a 2D helmet at the sensor level employing a bandwidth proportional to the central frequency. Once the signal was identified and accepted, we then employed ‘dSPM' (Dale et al, Neuron, 2000) a conventional minimum norm-estimate (MNE) based technique that allows visualization of the data on the cortical surface weighted by signal power and statistical significance (See Figure). The frequency band was split in 4 equidistant points then averaged. Results: sECD and dSPM analysis provided acceptable results in 8 and 9 out of 12 cases respectively. sECD and dSPM, when feasible, showed comparable results in terms of sub-lobar localization. There were two cases where dSPM provided superior localizing information as compared to sECD. Resection of the area of dSPM activation led to seizure freedom in both. sECD was not feasible in one case, while it localized to an area that was not resected in the second case. Resection of the area that contained the ictal sECD dipole appeared to correlate with Engel class Ia surgical outcome but the correlation was not statistically significant. On the other hand, this correlation was significant (RR 4.5 - 95% 1.33-15.28) when dSPM was employed. Conclusions: The results of this study suggest that dSPM analysis of a narrow frequency band of interest at seizure onset may provide additional and clinically relevant localizing information in patients with seizure onsets consisting of paroxysmal fast. Prospective and blinded studies will be needed to confirm the findings of this pilot study.