Performance of Automated Detection Software in Absence Seizures

Area of Science:

• Neurology
• Pediatric Epilepsy
• Medical Technology

Background:

• Absence seizures are frequent in children with idiopathic generalized epilepsies.
• Automated detection of absence seizures in electroencephalography (EEG) may improve efficiency.

Purpose of the Study:

• To evaluate the performance of Persyst software in detecting absence seizures in pediatric EEGs.
• To compare two detection modules: seizure detection and spikeburst detection.

Main Methods:

• Assessed Persyst's detection of absence seizures (≥3 seconds) in prolonged pediatric EEGs.
• Utilized seizure detection (most sensitive settings) and spikeburst detection (≥1 spike/sec, 1-3 sec duration).

Main Results:

• Seizure detection sensitivity was 78%; spikeburst detection sensitivity ranged from 61% to 95% based on duration.
• Combined module detection reached 99% for absences >9 seconds.
• False positive rates varied, with the 3-second spikeburst module yielding zero median false positives.

Conclusions:

• The spikeburst module with a 2-second duration threshold provides the optimal balance between sensitivity and clinical utility.
• Automated EEG analysis shows promise for identifying pediatric absence seizures.