s Wavelet Transform Based Algorithm for Automatic Detection of Patient- Specific Single Spike-and-Wave Discharges in Simulated Real-Time Conditions

Background: Transcranial magnetic stimulation applied at the appearance of spike-and-wave discharges in patients’ electroencephalograms may inhibit seizures. The prospect of transcranial magnetic stimulation holds much promise as a noninvasive treatment method for epileptic seizures, and the development of a system for the automatic detection of spike-and-wave discharges would facilitate implementation of this treatment method. However, the variety of waveforms and the appearance in the electroencephalography signal of waveforms similar to spike-and-wave discharges, called pseudo-spikeand- wave discharges, makes successful detection difficult to achieve. Objective: The aim of the current research was to develop an algorithm for the online detection of spikeand- wave discharges in epileptic patients’ electroencephalograms. Methods: In this study, a wavelet transform was used as the backbone for the algorithm. A clinician extracted data from a thirty-minute four-lead electroencephalography data recording, comprising fifty-four spike-and-wave discharge samples and fifteen pseudo-spike-and-wave discharge samples. Results: The simulated online detection method distinguished spike-and-wave discharges from pseudospike- and-wave discharges. However, a few cases of over-detection occurred, which has implications for the specificity and safety of the developed algorithm. Conclusion: The performance of a newly developed algorithm was reported. A visual analysis of the spike-and-wave discharges and pseudo-waveforms, as well as a time-frequency domain analysis, revealed features that make optimal detection of spike-and-wave discharge waveforms from other oscillations in electroencephalography recordings possible at a threshold level.