Our purpose in this study was to investigate the possible correlation of epileptic activity quantified by standard human methods and by a neural network trained to recognize epileptiform EEG patterns. We analyzed EEG recordings from 20 patients with a definitive diagnosis of generalized idiopathic or cryptogenic epilepsy. For each A/EEG recording the following variables were assessed: number, duration and circadian distribution of the epileptic discharges, electroclinical correlation, and median time for human and computerized net analysis. Human and computerized net analysis yielded more than 90% agreement (with an error less than 7%). The time required by the two physicians for quantifying epileptic discharges ranged from 90′ to 120′ while the net took an average 60′ and it could be used at any time of day without disturbing the general routine.
|Translated title of the contribution||Human operators vs a neural network for quantifying interictal EEG discharges in patients with epilepsy|
|Number of pages||2|
|Journal||Bollettino - Lega Italiana contro l'Epilessia|
|Publication status||Published - Jul 2006|
ASJC Scopus subject areas
- Clinical Neurology