Applications and comparison of pattern recognition from EEG and ECG signals

Traditional polisomnograph technique, used in assessing any statements related with sleep, has several disadvantages due to its complexness and cost. Currently newer automatic diagnostic technique are studied in diagnose and assessment as an alternative to it.One of the main aims of this study is presenting an automatic pattern recognition system for estimation vigilance level by using Electroencephalogram (EEG) signals recorded during transition from alert to sleep cases. In this system two different feature extraction methods based on Wavelet transform (WT) were used to detect the characteristics of EEG signals. Forcefulness of each feature extraction methods were tested on Artificial Neural Network (ANN), Least Squares Support Vector Machine (LS-SVM) and Adaptive neuro-fuzzy Inference System (ANFIS) classifiers. Also, performance comparison of six different classification approaches used for detection of vigilance level was done.The other main aim of this thesis is to present an automatic pattern recognition system for the automatic recognition of patients with Obstructive Sleep apnea syndrome (OSAS) from nocturnal Electrocardiogram (ECG) recordings. In this system, Heart rate variability (HRV), and ECG derived respiration (EDR) signals were obtained using an algorithm based on WT. Then two different feature extraction methods were used to determine the characteristics of both HRV and EDR signals. In the classification section of this system two different classifiers, which were ANN, and LS-SVM, were used. Twelve different classification approaches were done for detecting the situation of patients whether they have OSAS or not from ECG recordings by using this automatic pattern recognition system. In addition all the test performance results of these classification approaches were compared.Consequently, the systems presented in this thesis were compared with similar studies in the literature from the viewpoint of performance.Key Words: Vigilance Level, Obstructive Sleep Apnoea Syndrome (OSAS), EEG, ECG, Heart Rate Variability (HRV), ECG-derived Respiration (EDR), Pattern Recognition, Wavelet Transform (WT), Least Squares Support Vector Machines (LS-SVM), Artificial Neural Network (ANN), Adaptive Neuro-Fuzzy Inference Systems (ANFIS)