The ENCEVIS algorithm in the EMU and the factors affecting its performance: Our experience

Area of Science:

• Clinical Neurophysiology
• Medical Device Technology
• Epilepsy Research

Background:

• Long-term video electroencephalography (EEG) monitoring is crucial for epilepsy diagnosis and management.
• Automated seizure detection systems aim to improve efficiency and accuracy in analyzing EEG data.
• Evaluating the performance of new seizure detection algorithms is essential for clinical adoption.

Purpose of the Study:

• To assess the seizure detection performance of the ENCEVIS 1.7 algorithm.
• To identify factors influencing ENCEVIS 1.7's performance in detecting epileptic seizures.
• To explore the potential utility of ENCEVIS 1.7 in long-term video EEG monitoring units.

Main Methods:

• Analysis of 43 video-EEG recordings containing 112 epileptic seizures.
• Defined and calculated true positive, false negative, and false positive seizure detections.
• Investigated the influence of ictal pattern rhythmicity, seizure duration, patient age, and extracerebral signals on algorithm sensitivity.

Main Results:

• ENCEVIS 1.7 demonstrated an overall sensitivity of 71.2%, with higher sensitivity for focal (75.1%) versus generalized seizures (62%).
• Algorithm performance was influenced by rhythmic ictal patterns, longer seizure duration (>60 sec), and adult patient age (>18 years).
• ENCEVIS achieved 79.1% accuracy in annotating recordings with at least one seizure, showing reasonable performance for specific seizure types.

Conclusions:

• ENCEVIS 1.7 offers reasonable seizure detection capabilities, particularly for focal to bilateral tonic-clonic and temporal lobe onset seizures.
• Factors such as rhythmic ictal patterns, extended seizure duration, and adult age enhance algorithm performance.
• ENCEVIS can serve as a valuable tool for flagging seizure-containing recordings, potentially decreasing neurophysiologist workload.