Computer Assisted Automatic Sleep Scoring System Using Relative Wavelet Energy Based Neuro Fuzzy Model

This paper addresses the automated scoring of sleep stages using Electroencephalograph (EEG). The change in the Sleep Stages is accompanied by changes in the frequency spectrum of the EEG signals. A novel method based on Relative Wavelet Energy based Neuro-fuzzy is proposed to perform automatic sleep stages classification. Features extracted from 30-second epoch of (EEG) using relative wavelet energy are used for representing the EEG signal of different sleep stages. This method gives the best feature vector in terms of specificity and dimension. A neuro-fuzzy based ANFIS model is employed to classify these features to one appropriate stage. The sleep scoring is done for five stages namely, wake, sleep stages: stage1, stage 2, slow wave sleep (stage 3 & 4) and stage 5.The physionet database is used to validate the accuracy of the proposed automatic classification system. The hypnogram generated is compared with the standard hypnogram based on expert rule. The system can be used for real time implementation owing to high classification rate (97.4%), low computational cost, high speed and its feasibility to be implemented on hardware. The result of the study provides a framework of methodology that can be used to design computer assisted sleep scoring systems. Key-Words: Automated Sleep Scoring, hypnogram, EEG, Relative Wavelet energy, ANFIS, Physionet