Abstracts

USE OF A CLOUD-BASED EEG DATABASE TO MEASURE THE INTER-RATER RELIABILITY OF INTRACRANIAL EEG INTERPRETATION

Abstract number : 3.062
Submission category : 1. Translational Research: 1D. Devices, Technologies, Stem Cells
Year : 2013
Submission ID : 1751514
Source : www.aesnet.org
Presentation date : 12/7/2013 12:00:00 AM
Published date : Dec 5, 2013, 06:00 AM

Authors :
B. Oommen, J. Wagenaar, B. Litt

Rationale: The International Epilepsy Electrophysiology Portal (IEEG-Portal) is an online, cloud-based platform for sharing EEG data and tools among researchers. One goal of this platform is to provide high-quality EEG data that have been accurately annotated by experts trained in EEG interpretation. However, the reliability of EEG interpretation is variable. While the inter-rater reliability of human experts reviewing scalp EEG has been investigated, that of intracranial EEG (iEEG) is unknown. To explore this we used the IEEG-Portal to measure the inter-rater reliability of iEEG interpretation.Methods: The IEEG-Portal contains continuous iEEG data from humans who underwent presurgical evaluation. These recordings can be viewed and annotated via a viewer that runs in a web-browser. Selected epochs of iEEG recordings from subdural and temporal depth electrodes from 3 patients were reviewed by 2 epileptologists who marked seizure onset and offset times, and the electrode contacts in which each seizure was first observed. The degree of agreement between the reviewers for seizure duration was evaluated using intraclass correlation coefficient (ICC).Results: Twenty-four seizures were reviewed. Inter-rater agreement for seizure duration was fair (ICC 0.35, 95% CI -0.10-0.68). The electrode contacts identified by the two reviewers as the seizure origin overlapped in 94% of the seizures and were identical in 21%.Conclusions: The inter-rater agreement for interpretation of iEEG was fair for seizure duration, and localization of seizure onset was similar between the reviewers. These results suggest that more than one reviewer may be required for accurate iEEG interpretation. An online, cloud-based EEG data platform provides a convenient method for comparing iEEG annotation, and offers potential for collaborative review that may improve iEEG analysis.
Translational Research