A Theory for Multiresolution Signal Decomposition: The Wavelet Representation
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
The EEG is an affordable and non-invasive technique to study the brain and very useful in sleep analysis. The organization of the sleep in stages (global view) was widely used. The sleep EEG microstructure (local view) gives attention to the short duration EEG events. The CAPS (Cyclic Alternating Pattern Sequences) is a periodic EEG activity of sleep, and provides important information on EEG synchrony modulation in the sleep process and is closely related with the dynamic organization of sleep. It is characterized by repeated spontaneous EEG activations, at intervals up to one minute, from the background activity. The objective of this work is the automatic detection and classification of CAPS in sleep EEG. We use wavelet transforms as a tool to analyze the sleep EEG signal in the time-frequency domain, to separate the signal power in frequency bands. The CAP activation period (A phases) characteristics are used to build a detector and a state machine determines the CAP sequences periods. A group of sleep registers are tested and results compared with visual classification. A genetic algorithm is used to make the tuning of the detector.