High inter-reviewer variability of spike detection on intracranial EEG addressed by an automated multi-channel algorithm

Abstract

Objectives The goal of this study was to determine the consistency of human reviewer spike detection and then develop a computer algorithm to make the intracranial spike detection process more objective and reliable. Methods Three human reviewers marked interictal spikes on samples of intracranial EEGs from 10 patients. The sensitivity, precision and agreement in channel ranking by activity were calculated between reviewers. A computer algorithm was developed to parallel the way human reviewers detect spikes by first identifying all potential spikes on each channel using frequency filtering and then block scaling all channels at the same time in order to exclude potential spikes that fall below an amplitude and slope threshold. Its performance was compared to the human reviewers on the same set of patients. Results Human reviewers showed surprisingly poor inter-reviewer agreement, but did broadly agree on the ranking of channels for spike activity. The computer algorithm performed as well as the human reviewers and did especially well at ranking channels from highest to lowest spike frequency. Conclusions Our algorithm showed good agreement with the different human reviewers, even though they demonstrated different criteria for what constitutes a ‘spike’ and performed especially well at the clinically important task of ranking channels by spike activity. Significance An automated, objective method to detect interictal spikes on intracranial recordings will improve both research and the surgical management of epilepsy patients.

Intracranial EEG analysis in tumor-related epilepsy: Evidence of distant epileptic abnormalities

“…A validated spike detection algorithm was used initially to mark the interictal spikes of each 10-minute iEEG sample (Barkmeier et al, 2012b). Obvious artifacts were removed from the data by manually reviewing marked spikes.…”

“…These samples were then processed with our validated interictal spike detection algorithm (see below) (Barkmeier et al, 2012b). All surgical procedures were performed by the same epilepsy/brain tumor neurosurgeon (SM).…”

Interictal attenuation in pediatric electrocorticography can be reliably detected by EEG readers

“…Agreement between two EEGers independently assessing attenuation was moderate for the overall group (kappa = 0.56) and substantial for the physiologic In the ECoG literature, inter-rater agreements for interictal ECoG spikes were poor for identifying individual spikes, but the agreements were good for identifying the channels or regions of the brain with maximal spikes (Barkmeier et al, 2012;Dumpelmann and Elger, 1999). A direct comparison of their results to ours was not suitable given the differences in the way agreement was assessed (kappa vs. % in common vs. Kendall's coefficient of concordance, etc.)…”

Epileptic event detection algorithm for ambulatory monitoring platforms

“…Automated epileptic event detection has been studied with different approaches. Among the previously reported studies, some try to mimic human observers [7], others implement amplitude and frequency analysis [6,[8][9][10], frequency analysis with artificial neural networks [11], frequency and amplitude analysis through wavelets and machine learning algorithms [12,13], and finally decisions systems based on rules [14]. One of the most implemented and commercially used epilepsy event detection methods is the Gotman's algorithm [14].…”

“…Epileptic events have been described into four major groups: focal ictal patterns; focal inter-ictal patterns, generalized ictal patterns and generalized inter-ictal patterns. Focal ictal and inter-ictal events are more difficult to detect due to their high spatial, morphologic and inter-subject variability, these being the predominant factors to the poor inter-reviewer agreement [6]. Automated epileptic event detection has been studied with different approaches.…”