Automatic burst detection in a large cohort of extremely preterm infants, relation with brain maturation and neuro-developmental outcome

Abstract

Background The early identification of preterm infants at high neurological risk is essential. EEG is the only functional tool, which allows evaluating the cerebral maturation during the neonatal period. One of the EEG neuro-biomarker is the length Burst Activity (BA) adjusted to Post-Menstrual Age (PMA). For extremely preterms, studies about the predictive value of EEG are scarce, based on visual analysis in small cohorts. Objectives The objective of our study was to characterize BA in a large cohort of extreme preterms, using an automated burst detection tool develop and optimize in our laboratory, and to correlate BA characteristics with the brain maturation and the neurological outcome. Methods A total number of 86 preterm infants born before 28 weeks GA at Amiens University Hospital were included. Outcome assessment was based on Denver Developmental Screening Test Scale, at 2 years. The first EEG of each infants recording before 30 weeks PMA was analyzed with the automated detection method. Results The mean BA duration was significantly higher for infants with favorable outcome compared to infants with unfavorable outcome. With the PMA, the cerebral maturation of infants with the favorable outcome is characterized by an increase of the mean BA duration associated with a decrease of the number of BA. Conversely, in case of unfavorable evolution, the mean BA duration remains below 10 s despite the increase of PMA. Conclusion Automatic BA tool allow evaluating the brain maturation in extremely preterms and defining objective neurobiomarker of neurodevelopmental outcome.