Abstracts

Rationale: Magnetoencephalography (MEG) is a powerful tool to estimate the current source of epileptic discharge. However, estimation of a deep-seated focus is challenging because of its low signal-to-noise ratio and the ambiguity of current sources estimated by respective interictal epileptiform discharge (IED). In this study, we used modified distributed source (DS) analysis, using a volume head model as the source space of the forward model, in combination with statistical methods to reduce variation. We also evaluated the efficacy of modified DS analysis in surgical cases. Methods: We retrospectively analyzed 19 cases with focal epilepsy that were preoperatively evaluated by MEG and subsequently underwent surgery. In the modified DS analysis, we used a volume head model constructed from each patient’s entire brain, in combination with standardized low-resolution brain electromagnetic tomography and statistical methods, such as permutation tests between IEDs and baselines and false discovery rate between voxels. Furthermore, we performed equivalent current dipole (ECD) analysis. Surgically resected areas were evaluated and the concordance rates of identified epileptic focus were calculated. These results were compared with those of surgical outcomes. Results: The concordance rates from the modified DS analysis were significantly higher than those from ECD analysis (68.4% vs. 26.3% in all cases), especially in cases with deep-seated lesions such as the cingulate gyrus and mesial temporal structures (81.8% vs. 9.1%). Modified DS analysis was also helpful in differentiating mesial from neocortical temporal lobe epilepsy. Furthermore, the concordance rate correlated well with surgical outcomes. Conclusions: We succeeded in identifying regions with statistically significant current distribution of the IED recorded by MEG localization, using the volume head model as the forward model, and statistical methods applied to sLORETA of the DS analysis as the inverse model. The concordance rates from this modified DS analysis were significantly higher than those of ECD analysis, especially in cases with deep-seated lesions such as in the cingulate gyrus or mesial temporal structures. It was also helpful in differentiating mesial from neocortical temporal lobe epilepsy. Furthermore, the concordance rate correlated well with surgical outcomes. Modified DS analysis can improve diagnostic performance for deep-seated epileptic foci, and allow more accurate and less invasive determination of the surgical resection area. Funding: This work was supported by JSPS KAKENHI Grant Number 17K10890 (principal investigator; Satoshi Maesawa).