ANALYSIS AND CLASSIFICATION OF BRAIN SIGNALS USING DWT AND SVM

摘要

The brain is the main dominant of the human body and any defect that occurs in the brain impact the body's vital activity. There are many diseases that affect on the brain tasks. They Infects many people worldwide. The Brain signals recorded by Electroencephalography (EEG) system are used to diagnose brain signals and classify them as normal or abnormal. Detecting and classifying of EEG signals are exhausting and difficult process and require effort by the neurologist to diagnose them. When the brain injury is severe, treatment of brain injuries become difficult and require surgical intervention, but in early detection of the injury it may be treated without surgical intervention. The purpose of this paper is to detection and analysis of the brain signals and extract the main bands of the signal by Discrete Wavelet Transform (DWT) and then to classify them by the Support Vector Machine (SVM) for early detection of brain abnormalities. The database used in this paper were collected from a group of patients at Baghdad Teaching Hospital / Medicine City by an EEG system in the form of images. The signals are extracted from the images by detecting the signal inside the image and dealing with it as a digital signal. These signals are immersed by noises to eliminate this noise the Finite Impulse Response (FIR) filter is used where low frequencies pass and block the high frequencies to obtain noise-free signals. Set of statistical measurements are measured to use as input to train the Support Vector Machine and used them in classification of the signal. The classification ratio obtained in this paper is 96.8 %.