Adaptive separation of background activity and transient phenomenon in epileptic EEG using mathematical morphology and wavelet transforms

Epileptic EEG data contains transient activities, such as spikes, muscle activities, eye movements and artifacts. The transients are complex in shape, occurring randomly and with short duration. The background EEG often appears as a slow wave in contrast with transients. In the past, many methods have been investigated to separate the two components of background activity and transients. These methods detect background activity or localize the transients by using linear signal processing techniques, which are not effective for detecting sharp waves and patterns. This paper presents a new approach to signal separation by directly judging the differences between the components' shapes using morphological analysis. Morphological analysis utilizes analytic operations based on a pre-defined structuring element (SE). These operations, implemented by a filter, can detect and extract specific signal features determined by the SE. In our case, the SE is defined as a circle to measure the degree of smoothness between the two signal components. A discrete wavelet transform is applied to construct the processed signal. The multi-resolution property of the wavelet transform adapts well to the time-invariant nature of the signal. Combining mathematical morphology and wavelet transforms, this method has successfully separates the background activity and transient phenomenon from an epileptic EEG signal.