Abstracts

Diagnostic Accuracy of Interictal Electric Source Imaging in the Presurgical Evaluation of Patients with Drug-Resistant Focal Epilepsy

Abstract number : 1.168
Submission category : 3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year : 2023
Submission ID : 121
Source : www.aesnet.org
Presentation date : 12/2/2023 12:00:00 AM
Published date :

Authors :
First Author: Rafael Toledano, MD – Hospital Ruber Internacional; Hospital Ramón y Cajal. Madrid, Spain

Presenting Author: Irene Sanchez-Miranda Roman, MD – Hospital Ruber Internacional, Madrid, Spain

Paloma Parra-Díaz, MD – Hospital Ruber Internacional; Hospital Ramón y Cajal. Madrid, Spain; Adrián Valls-Carbó, MD – Hospital Ruber Internacional. Madrid, Spain; Álvaro Beltrán-Corbellini, MD – Hospital Ruber Internacional. Madrid, Spain; Irene García-Morales, MD, PhD – Hospital Ruber Internacional; Hospital Clínico San Carlos. Madrid, Spain; Jaime Masjuan, MD, PhD – Hospital Ramón y Cajal. Madrid, Spain; Antonio Gil-Nagel, MD, PhD – Hospital Ruber Internacional. Madrid, Spain

Rationale:
Electric source imaging (ESI) of interictal epileptiform discharges (IEDs) has been shown in previous studies to be a highly useful tool for localizing the epileptic focus in the presurgical evaluation of patients with drug-resistant epilepsy (DRE). However, it is not widely used in many centers as it requires prior learning and validation. In addition, various source models can be applied to calculate the inverse solution, although comparative studies are lacking. Our objective is to analyze the diagnostic accuracy of multiple inverse solutions performed with an open-source software (Brainstorm).



Methods:
A retrospective analysis was conducted to evaluate the diagnostic accuracy of six different source models for ESI using Brainstorm. Patients with DRE who had undergone epilepsy surgery and had at least one year of follow-up, in whom a presurgical prolonged video-EEG (> 25 electrodes) and a postoperative brain MRI had been performed, were included. The methods used to calculate the inverse solutions were Minimum Norm Imaging (current density maps, dSPM and sLORETA models), LCMV beamformer, Maximum Entropy on the Mean and dipole fitting. Specificity, sensitivity, predictive values, accuracy and odds ratio were calculated for each solution at both the peak and 50% of the ascending curve of the averaged spikes. 


Results:
A total of 82 patients were included, of whom 63.4% had extratemporal epilepsy, 67.1% had a focal lesion identified on MRI, 57.3% underwent stereo-EEG (SEEG) prior to surgery and 72% were classified as Engel I at one-year-follow-up. A total of 984 inverse solutions were analyzed, with a mean of 102 averaged IEDs per patient. Overall, the inverse solutions obtained at 50% of the ascending curve were found to have higher diagnostic accuracy in all the evaluated models. Among them, sLORETA was observed to be the most sensitive (Se = 80%) and specific (Sp = 74%), with an accuracy of 78% and an OR of 11.1 for achieving a seizure-free outcome if the identified volume was included in the surgical resection. The diagnostic accuracy of sLORETA remained high in patients with extratemporal epilepsy (Se = 84%, Sp = 80%, accuracy = 83%), non-lesional MRI (Se = 92%, Sp = 67%, accuracy = 78%), and those who required SEEG (Se = 83%, Sp = 67%, accuracy = 77%). Sensitivity was significantly higher (p < 0.001) for sLORETA and dipole fitting performed at 50% of the ascending curve in cases of non-lesional MRI and prior SEEG.
Neurophysiology