Abstracts

Rationale: Paediatric epilepsy surgery is increasingly important due to the potential benefits of earlier surgery to optimise seizure and cognitive outcome (Skirrow, Cross et al. 2011). In this population EEG is often less localising, IEDs are more frequent and less-invasive test with good sensitivity and specificity are needed. EEG-fMRI maps often show multiple regions of activity which, while being consistent with the idea that epilepsy is a network disease (Centeno and Carmichael 2014), complicates interpretation where a single spatial target is typically required for surgery to proceed. ESI has been shown to localize the ER with high sensitivity and specificity in a mixed group of adults and children using high density (>32 channel) EEG (hdEEG) (Brodbeck, Spinelli et al. 2011) and shown promise in a paediatric cohort (Russo, Jayakar et al. 2016) albeit with low density ( < 32 channels) EEG. The combination of both test has been shown to reveal complementary information and this information can be combined to mutually inform both maps (Vulliemoz, Lemieux et al. 2010). In this study we propose to take advantage of hdEEG-fMRI data to derive maps from both ESI and IED-correlated fMRI. We aimed to assess the localising and predictive value of EEG-fMRI and ESI individually and as a combined test in the largest paediatric focal epilepsy population studied to date with either method. Methods: Fifty three children with drug-resistant epilepsy and frequent IEDs underwent EEG-fMRI using a child centred EEG-fMRI protocol (Centeno, Tierney et al. 2016). Twenty patients subsequently underwent surgery. Two type of analysis were performed 1) EEG-fMRI (either using a model of interictal epileptiform discharges (IED) or topographic voltage map correlation analysis (Grouiller, Thornton et al. 2011)); 2) Electrical Source Imaging (ESI)(Brodbeck, Spinelli et al. 2011) . A single localization was derived by following the procedure seen in Figure 1.The contribution of combined EEG-fMRI and ESI to characterize the epileptogenic region more accurately was quantified. Accuracy measures (sensitivity, specificity, positive and negative predictive values) to predict surgery outcome were calculated for the localization obtain by this method and compared to the localization of EEG-fMRI and ESI alone. Results: Fifty two of 53 patients had significant results, 34/53 for both EEG-fMRI and ESI, 13/53 for EEG-fMRI only and 5/53 for ESI only. A single localization was derived with the described method in 47/53. This localization contributed to a more spatially specific characterization of the epileptic focus as defined clinically in 64% of all patients and in 82% of MRI negative cases. 19/47 had surgery and the localisation predicted outcome with 100% sensitivity and 83% specificity. In contrast, EEG-fMRI alone and ESI alone had lower sensitivity and specificity scores (21%/83% and 83%/75% respectively). A concordant EEG-fMRI and ESI cluster was found in 22/47 and predicted outcome with 100% sensitivity and specificity. Conclusions: EEG-fMRI combined with ESI has a high localization accuracy and is particularly useful in in providing a spatial hypothesis in MRI negative cases. Funding: Action Medical Research