Using spectral frequency analysis and visualization techniques on pediatric electroencephalogram recordings, we identify a method of detecting seizures thirty seconds before seizure onset by identifying significant preictal locations and their respective frequencies within the gamma band of 30 through 100 Hz. Using average log-power differences and visualization techniques to identify significant preictal patterns, intractable seizures are found to have common frequency extremes (CFEs) in the high gamma band between 70 and 100 Hz. Using this data, machine learning detection algorithm predictive performance may be improved by incorporating high gamma band signal processing at the varying location(s) and strength of a pediatric patient’s CFEs.
DOI: 10.1109/ACCESS.2021.3087782