Search

Seizures occur in the neonatal period more frequently than other periods of life and usually denote the presence of serious brain dysfunction. The gold standard for detecting seizures is based on visual inspection of continuous electroencephalogram (cEEG) complemented by video analysis, performed by an expert clinical neurophysiologist. Previous studies have reported varying degree of agreement between expert EEG readers, with kappa coefficients ranging from 0.4 to 0.85, calling into question the validity of visual scoring. This variability in visual scoring of neonatal seizures may be due to factors such as reader expertise and the nature of expressed patterns. One of the possible reasons for low inter-rater agreement is the absence of any benchmark for the EEG readers to be able to compare their opinions. One way to develop this is to use a shared multi-center neonatal seizure database and use the inputs from multiple experts. This will also improve the teaching of trainees, and help to avoid potential bias from a single expert's opinion. In this paper, we introduce and explain the NeoGuard public learning platform that can be used by trainees, tutors, and expert EEG readers who are interested to test their knowledge and learn from neonatal EEG-polygraphic segments scored by several expert EEG readers. For this platform, 1919 clinically relevant segments, totaling 280h, recorded from 71 term neonates in two centers, including a wide variety of seizures and artifacts were used. These segments were scored by 4 EEG readers from three different centers. Users of this platform can score an arbitrary number of segments and then test their scoring with the experts' opinions. The kappa and joint probability of agreement, is then shown as inter-rater agreement metrics between the user and each of the experts. The platform is publicly available at the NeoGuard website (www.neoguard.net).