Automatic Detector of Abnormal EEG for Preterm Infants

TitreAutomatic Detector of Abnormal EEG for Preterm Infants
Type de publicationCommunications avec actes
Année de publication2017
Titre de la Conférence/colloqueInternational Conference on IoT Technologies for HealthCare
jour/mois du congrès, colloque09/2017
Auteur(s)Jrad, N., Schang D., Chauvet P., Daya B. et Gibaud M.
Université, EditeurSpringer
Ville, PaysAngers, France
Numéro ISBN978-3-319-76213-5
Mots-clésAutomatic EEG analysis, Feature extraction, Inter Burst Interval Detection, Multiple Linear Regression, Preterm infants

Many of preterm babies suffer from neural disorders caused by birth complications. Hence, early prediction of neural disorders, in preterm infants, is extremely crucial for neuroprotective intervention. In this scope, the goal of this research was to propose an automatic way to study preterm babies Electroencephalograms (EEG). EEG were preprocessed and a time series of standard deviation was computed. These series were thresholded to detect Inter Burst Intervals (IBI). Features were extracted from bursts and IBI and were then classified as Abnormal or Normal using a Multiple Linear Regression. The method was successfully validated on a corpus of 100 infants with no early indication of brain injury. It was also implemented with a user-friendly interface using Java.