Abstracts

Rationale: The purpose of this study was to examine the concordance of localizing information between magnetic source imaging (MSI) and electrical source imaging (ESI) acquired concurrently during the same study.  As MSI is more sensitive to sulcal surfaces and ESI more sensitive to surface convexities, it was hypothesized that there should be some degree of difference in localizations. Methods: A database of studies at Texas Children's Hospital was reviewed to identify studies where ESI was done concurrently with MSI.  A total of 88 studies were identified.  The MEG data was acquired using the Elekta Neuromag Triux system.  EEG was recorded using standard 10-20 electrode positions, plus subtemporal electrodes.  At least one hour of spontaneous combined recording was collected in each case.  Both MSI and ESI was accomplished using single equivalent current dipole models and co-registered to the patient's MRI.  Elekta's proprietary software and Compumedics Curry 7 were used.  All studies were categorized into four different categories: "Discordant" when it was felt that the ESI and MSI results gave different localizations; "Concordant" when it was felt that the ESI and MSI results matched and neither gave additional localizing information; "Concordant/MSI+" when it was felt that both data sets were concordant, but the MSI results gave additional localizing information; and "Concordant/ESI+" when was felt that the data sets were concordant, but the ESI results have additional localizing information. Results: Of the 88 combined MSI/ESI studies, the percentage of suspected frontal, temporal, parietal, and occipital lobe cases were 31.8%, 20.5%, 8%, and 4.5%, respectively.  A further 3.4% were felt to be right hemispheric onset, 2.2% left hemispheric onset, and 2.2% were felt to be due to hypothalamic hamartomas.  19.3% were felt to be multifocal, involving multiple lobes, and 7.9% were felt to be of unclear onset.  As a whole, 10 cases were felt to be completely discordant (11.3%), with the ESI and MSI data giving different information.  The remainder (88.7%) were felt to fall into one of the concordant categories.  In 35 cases (39.7%) both data sets were felt to be completely concordant.  Interestingly, in 34 cases (38.6%) it was felt that the MSI data gave extra localizing information, but in another 17% it was felt that the ESI data gave extra localizing information that was not given by the MSI data.  This was mostly due to spikes that were clearly seen in EEG, but not MEG (7 cases), with most of these cases being frontal or frontopolar.  Other reasons included spike propagation information that was only seen in EEG  (2 cases) and different sub-lobar classifications (5 cases).  Regarding extra localizing information identified in Concordant/MSI+ cases, the vast majority of the additional information was additional sublobar localizations (12 cases) and identification of MEG only spikes (12 cases).  One case was due to additional spike propagation information not seen in the EEG data. Conclusions: The results of this study suggest that MSI has a high likelihood (34%) of providing additional localizing information not seen in the ESI data.  The more striking result, however, is that ESI also a significant likelihood (15%), although smaller, of providing localizing information not seen in the MSI data.  In both groups (Concordant/MSI+ and Concordant/ESI+), the reasons were largely for epileptiform spikes not seen in the other modality, and for additional sublobar localizations not seen in the other modality.  This finding is significant, as it is not common in most clinical MEG centers to perform ESI, and may be an argument for ESI implementation as a standard of practice. Funding: None