Current Proceedings on Technology
Yazarlar: Baha Şen, Musa Peker
Konular:-
Anahtar Kelimeler:EEG signals,Classification,Feature selection,Sleep stage
Özet: Classification of sleep stages is one of the important methods for diagnosis in psychiatry and neurology. Sleep scoring is a time-consuming and difficult task performed by sleep specialists. This study proposes an efficient approach for neurologists to help them to classification of sleep stages with high accuracy rates. This study consists of three stages: feature extraction, feature selection and classification stage. In the feature extraction stage, we used 20 attribute algorithms in four categories. 41 feature parameters were obtained from these algorithms. In feature selection phase, Minimum Redundancy Maximum Relevance (mRMR) feature selection algorithm is chosen to select a set of attributes that better represent the EEG signals. The resulting attributes are used as input parameters for decision tree algorithm. The evolution of the proposed system is conducted using k-fold cross-validation, classification accuracy, sensitivity, specificity values and confusion matrix. The results are promising and a classification accuracy of 92.3% is achieved.