Automatic detection of generalized paroxysmal fast activity in interictal EEG using time-frequency analysis

Objective Markup of generalized interictal epileptiform discharges (IEDs) on EEG is an important step in the diagnosis and characterization epilepsy. However, manual markup a time-consuming, subjective, specialized task where human reviewer needs to visually inspect large amount data facilitate accurate clinical decisions. In this study, we aimed develop framework for automated detection paroxysmal fast activity (GPFA), IED seen scalp recordings patients with severe epilepsy Lennox-Gastaut syndrome (LGS). Methods We studied 13 children LGS who had GPFA events their recordings. Time-frequency information derived from manually marked IEDs across multiple channels was used automatically detect similar each patient's EEG. validated true positives false proposed spike approach using both standalone simultaneous EEG-functional MRI (EEG-fMRI) Results displayed consistent low-high frequency arrangement time-frequency domain. This 'bimodal' spectral feature most prominent over frontal channels. Our automatic identified properties GPFAs. Brain maps EEG-fMRI signal change during these detected were comparable brain markup. Conclusion have characteristic bimodal that can be LGS. The validity demonstrated by analysis events, which recapitulates previously shown underlie Significance study provides novel methodology enables fast, automated, objective inspection may extendable wider range syndromes monitoring burden epileptic aid decision-making faster assessment treatment response estimation future seizure risk.