Feature Extraction from EEG using Wavelets
wavelet transforms with basic detection theory to develop a new unsupervised method for robustly detecting and localizing spikes in noisy neural recordings. The method does not require the construction of templates, or the supervised setting of thresholds, it is based on a Bayesian detector where the noise is modeled to be Gaussian in nature. We present simulations on actual EEG recordings under different recording conditions. We further demonstrate that falsely detected spikes corresponding to our method resemble actual spikes more than the false positives of other techniques such as amplitude threshold. Moreover, the simplicity of the method allows for nearly real-time execution. On discussions with neurologists specializing in sleep related disorders and epilepsy, we were made aware of the tediousness of diagnosing sleep related disorders and epilepsy.Diagnosis involves accurate detection and localization of the individual spikes and spike trains from EEG signals acquired from sleep tests, which are recorded for several hours. Recorded spike trains are inevitably corrupted by noise. The source(s) for the noise are highly varied. Perhaps most importantly, the activity of distant neurons may appear as noise which is highly correlated with the useful signal. Another difficulty is that the spikes (EEG spike trains/EEG spikes are used synonymously) both are have the shapes and amplitudes that are highly variable. All of these issues make the problem of spike detection challenging.