Detection of subclinical electroencephalographic seizure patterns with multichannel amplitude-integrated EEG in full-term neonates

Abstract

Objective: To compare the seizure pattern detection rate of single-channel and multichannel amplitude-integrated EEG (aEEG), using conventional EEG (cEEG) as a gold standard, in full-term neonates with hypoxic-ischemic encephalopathy. The optimal electrode derivation for seizure detection with single-channel aEEG was also investigated.

Methods: Twelve infants with cEEG seizure patterns (10s) were investigated. cEEG signals were transformed into aEEG signals. Seizure patterns and the number of patients identified with 1 seizure patterns were calculated for single- and multichannel aEEG.

Results: On cEEG, 121 seizure patterns with a mean duration of 58s were identified, 68% of which occurred over the centrotemporal region. The sensitivity of aEEG for the detection of seizure patterns was 30% (C.I.: 0.22-0.38) for single-channel aEEG and 39% (C.I.: 0.31-0.48) for multichannel aEEG. Multichannel aEEG identified all patients with 1 seizure pattern (C.I.: 0.75-1.00), whereas single-channel aEEG (with C4-C3 as the optimal electrode derivation) identified all but one of the patients (C.I.: 0.66-0.99).

Conclusions: Seizure pattern detection rate is slightly better with multichannel aEEG compared with single-channel (C4-C3) aEEG. Multichannel aEEG identified correctly all patients with 1 seizure pattern in this small selection of patients.

Significance: Single-channel aEEG may detect most patients (in a selected group) with severe neonatal seizures patterns; patient identification can be improved using multichannel recordings.