Abstracts

Rationale: EEG-fMRI is a noninvasive useful tool to detect epileptic generator for some patients with medically refractory epilepsy. As in most methods, results are not always significant and are not always in agreement with other localization approaches. We tried to determine if some factors influence the likelihood of having positive results. We hypothesized that the number of interictal epileptic discharges (IEDs) during the scan and the spatial extent of IEDs are contributing factors in obtaining appropriate activations that reveal the seizure onset zone. Methods: Seventy-one consecutive EEG-fMRI studies with significant BOLD activation (t-value >3.1) from patients aged 16 years or older with refractory localization-related epilepsy scanned from April 2015 to March 2018 in our institute were reviewed retrospectively. Sixty-four studies remained after excluding the 7 studies corresponding to a second scan of the same patient. In these 64 studies/patients, 157 IED types were evaluated. One IED type corresponded to one EEG scalp distribution. To determine reliable significant activation, we first thresholded the activation map of each type at a t-value > 3.1. We then identified the clusters with maximum t-value higher than a more conservative threshold, the significance threshold corresponding to corrected whole-brain topological false discovery rate (FDR) of 0.05.The location of the maximum activation was compared to the presumed seizure onset zone (SOZ) as defined by all tests available for the patient, including clinical symptomatology, video-EEG and MRI in all patients and other tests in some (PET, ictal SPECT, MEG). Results:  In 157 types of IEDs, t-values in 109 were above FDR, t-values in 43 types were above the 3.1 threshold but below FDR and 5 types had no significant positive activation. The number of IEDs was significantly higher in the types with t-value above FDR than in the types with t-value below FDR (p<0.01). Among the 109 IED types from 58 patients, 63 types occurred in the 30 patients in whom a presumed SOZ could be defined. In these 63 types, the location of maximum activation in 36 types was concordant with the presumed SOZ and it was discordant in the remaining 27. The types of IEDs concordant with the presumed SOZ had a significantly larger extent on scalp EEG than the types discordant with the presumed SOZ (p<0.01). Conclusions: The number of IEDs is an important factor in obtaining reliable significant activations in EEG-fMRI. In addition, IEDs with a larger spatial extent are more likely than focal IEDs to reveal an accurate location for seizure onset zone. This is counterintuitive but may result from the fact that widespread discharges require more energy than focal discharges to spread from their source (the spike onset zone) and therefore result in a statistically more reliable activation at the source. Funding: No funding was received in support of this abstract.