Modified expectation maximization based sparse representation classifier for classification of epilepsy from EEG signals

To control the coordination between various muscles and nerves in the human body, brain is utilized. Due to the sudden, unexpected and transient electrical disturbances of the brain, it results in an acute disease called epilepsy which is characterized by recurrent seizures leading to a lot of temporary changes in behaviour, perception, movement and health. Epilepsy occurs due to the rapid firing of neurons in the cortical sections of the brain. To record the electrical activities of the brain and to analyze the patients who are suffering from epilepsy, Electroencephalography (EEG) is widely used. As the recordings of the EEG signals are done for a long period of time, the dimensions of the EEG signal have to be mitigated so that classification procedures become easier. In this paper, Autoencoders dependent on Genetic Algorithm (GA) are used to reduce the dimensions of the EEG data. The dimensionally reduced values are then fed inside the classifier called Modified Expectation Maximization Based Sparse Representation Classifier (MEM-SRC) for the perfect classification of epilepsy from EEG signals. The result analysis is shown in terms of Performance Index, Quality Values, Time Delay, Specificity, Sensitivity and Accuracy.