Abstracts

Rationale:
Accurate localization of the epileptogenic zone (EZ) is the key to obtain good curative effect of surgery for refractory epilepsy patients.Over the past two decades, a number of new techniques have been developed to localize theEZ, including PET-MRI, special MRI sequences( such as fluid and white matter suppression, FLAWS), neuroimaging post processing technique, and automatic analysis of high frequency oscillations (HFOs) and so on. However, to date, no single technique has been able to precisely locate the EZ.In this study，we aim to explore the possibility of the multimodal method by jointly applying PET-MRI, FLAWS and HFOs automated analysis to delineate the EZ. We also want to find out the accuracy of different combinations to locate the EZ.
Method:
Eight epilepsy surgery patientswith at least two above examinations and beyond one-year postsurgical follow-up were included in this study. We compared the abnormal regions marked by PET-MRI, FLAWS, and HFOs in different combinations, and analyzed their overlapped areas (at least two or more methods) with removed volume (RV) and outcome. What’s more, RatioChann(ev) (i.e., channels inside versus outside the RV) was used as a nonspecific measure to evaluate how well the channels marked by three methodsare correlated with the RV. To ulteriorly investigate the correlationamong the three methods, every patient’s extent of marking areas for each method are also compared.
Results:
The consistency of marking suspicious areas between PET-MRI&HFOs, FLAWS&HFOs, PET-MRI&FLAWS were 42.86%, 33.33% and 80%, respectively. When the overlapped areas detected by two or more methods were removed, seizures after surgery were reduced by at least 90 percent. The RatioChann(ev)of HFOs marked contacts was significantly higher in patients with a good outcome than in those with a poor outcome (p=0.014). For PET-MRI and FLAWS, there were no significant difference about the ratios between patients with good and poor outcome (p=0.186, p=1). After comparing the extent of areas for each patient marked by the three methods, we found that in most cases, PET-MRI cover largest hypometabolism areas, FLAWS’s marking area seem to be smaller and HFOs meaningful areas are the most focal, which locate in the core or the margin of lesion or the perilesion marked by neuroimaging. We also found that the periphery of lesion (focal cortical dysplasia, FCD) marked by neuroimaging may be epileptic and not every lesion (especially for nodular heterotopia, NH)could contribute to epileptic seizure, so it wasmore meaningful to emphasize the combinationof these three methods to delineate EZ.
Conclusion:
Approaches in multimodality can detect EZ more accurately and HFOs analysis may help to define real epileptic areas that may be missed in the neuroimaging results. The implantation of intracranial electrodes guided by non-invasive PET-MRI and FLAWS findings as well as HFOs analysis would be an optimized multimodal approach for locating EZ.
Funding:
:This study was funded by National Natural Science Foundation of China (XFY 81971202, 81671367, 81790653, and 81471391) and the Beijing Natural Science Foundation (GJZ L182015).