Abstracts

The clinical outcome of neonatal seizures has been studied with some success. Combination of EEG, clinical findings and neuroimaging studies are useful predictor of developmental outcome, but do not correlate well with post neonatal seizures (PNS). We applied new frequency analysis method, Multiple Band Frequency Analysis (MBFA), for neonatal seizures. MBFA provides both high frequency and temporary resolution to analyze frequency components and time course of neonatal EEG seizures. We compared the MBFA findings between the group with PNS at the age of 18 month-old (Group A) and the other group without PNS (Group B).
We reviewed medical records of neonates admitted to the Hospital for Sick Children (HSC) from 1996 to1998, which documented electrographical seizures. We sifted the records that had clinical follow up at least 18 months at HSC. We selected ictal EEG that lasts more than 10 seconds. We used digital EEG system with 16 channels bipolar neonatal EEG and extracerebral monitoring channels, including EOG, ECG, EMG and thoracic (respiratory) movements. Digital EEGs are recorded at least 30 minutes at sampling rate of 200 Hz. We employed MBFA to analyze EEGs and to make power spectra. MBFA applied multi band-pass filter to decompose EEG signals to particular frequency component. MBFA measured amplitudes of decomposed frequencies in every assigned time window. MBFA provides power spectra with obtained amplitudes of every assigned frequency band. MBFA can trace the highest power frequency in every assigned time window. In this study, we analyzed the neonatal EEGs that range from 1 Hz to 20Hz with 0.2Hz frequency resolution in every 100mS. We collected duration, spectral power and frequency of the sustained dominant seizure activities that calculated by MBFA. We compared those data between Group A and Group B with Mann-Whitney[rsquo]s U test.
Twenty-eight patients presented neonatal seizures. We excluded 13 patients as they lacked post-neonatal follow-up or qualified EEG recordings. Group A had 11 neonates and Group B had 4 neonates. We obtained 52 electrographical seizures in Group A (2-7 seizures per patient, average 5 seizures) and 16 seizures in Group B (2-7 seizures per patient, average 4 seizures). Group A had the longer duration of the sustained peak frequency (mean [plusmn] SD; 9.5 [plusmn] 5.4 sec.) than that of Group B (4.8 [plusmn] 3.0 sec.) with a statistical significance (p[lt]0.01). There were not any significant differences in the peak frequency (2.0[plusmn] 1.5 Hz V.S. 1.8[plusmn] 0.9 Hz) and the power of the peak frequency (26.1[plusmn]18.1 dB V.S. 30.7[plusmn]13.2 dB) between two groups.
The neonates who had longer duration of sustained peak frequency developed PNS. MBFA is a useful tool to analyze frequency components and time course of neonatal EEG seizures for prediction of developing epilepsy in post neonatal period.