Abstract

Electroencephalogram (EEG) is a way of monitoring the brain's spontaneous electrical activity. It is considered to be the best way for diagnosing tumour and other brain disorders in the human brain. This work presents a diagnostic system to classify EEG signal as tumour or normal. Here the system uses the back propagation with a feed forward algorithm for classification, which uses the neural network classifier. During the neural network training phase, the extracted statistical features from the EEG signal were presented to the neural network with the help of set of database samples. During the neural network testing phase, the neural network classifies the test sample set as normal or brain tumour based on the training given. The obtained results show that, the proposed system gives better classification accuracy for the different test samples over the existing methods.

Introduction:

Signal processing plays a major role in the analysis of the different modes of signals, images, sounds and various biological measures. The processing of the biological signals is mainly concerned with the detection of various disorders related to the patient's physical nature. Thus, for the analysis of different states of the human brain we need to extract the brain signals with the help of electrodes and electroencephalogram. Extraction of the brain signals needs some sort of filters and wave metrics to work on the extracted signal input. We have to select a proper algorithm to process the signal and obtain the output for future detections. In this research paper, brain signal is extracted using electroencephalogram and further pre-processing techniques are performed to obtain the final classification of the brain tumour Electroencephalogram remains the primary reorganisation test of brain function. EEG is especially valuable in the investigation of patient's electrical activity produced by the cerebral cortex nerve cells and it is extensively used in clinical categorization of brain activities. The scalp may produce various electrical potentials which represent the variation in the brain's activity. Besides from various new imaging
techniques like MRI, SPECT and PET, EEG provides better diagnostics responds to the tumours. EEG signals are broadly classified into delta, theta, alpha, beta and gamma signals. While processing the EEG signals, the first part includes acquisition of signal, removal of disturbance or noise in the signal, thresholding, enhancement and finally detection of the signal. The final stage in EEG signal analysis includes selecting matching algorithms and techniques to define the brain's activity Analysis of the brain tumour by the physicians only records the neural activity of the brain. It is very important to detect the brain tumour in the human in its early stages. Mostly examining the signals of the brain is a time-consuming process, but it may be analysed by using accurate methods and visualizing of the signals. The presence of any external stimulus and disturbances in the brain leads to non-stationary EEG recordings.