Prediction of secondary generalization from a focal onset seizure in intracerebral EEG

Abstract

Objective: We propose a system based on the first five seconds of intracerebrally recorded focal seizures to predict their evolution from focal to bilateral tonic-clonic seizure (FTC), to spread outside the onset zone but without tonic-clonic component (FS), or to a seizure remaining focal (F).

Methods: Nineteen time and frequency domain features were extracted from 39 seizures of 32 patients and were subjected to support vector machine based classification. Three prediction approaches regarding seizure evolution were (1) FTC vs FS vs F, (2) FTC vs FS or F and (3) FTC or FS vs F.

Results: We used 21 seizures for training. Most features had significantly different distributions in the three seizure types (p < 0.05). Eighteen seizures were used for testing. We predicted the evolution of 12 seizures in FTC vs FS vs F, 13 seizures in FTC vs FS or F and 14 seizures in FTC or FS vs F.

Conclusion: The first five seconds of a focal seizure contains information regarding the eventual evolution of the seizure, which could be predicted in most seizures.

Significance: The proposed system could alert the health care team when a patient is hospitalized for intracerebral EEG and improve safety and eventually result in an implantable device.

Keywords: Automated system; Intracerebral EEG; Multiscale entropy; Relative power; Tonic-clonic seizure.