ANN based Detection and Diagnosis of Neurological Disorder Diseases
The electrical action of the brain is recorded by using the instrument called Electroencephalogram (EEG) which is the most useful device in the assessment of neurological disorders. Detection of virtual diagnostics is time-consuming and difficult work and therefore in this paper we propose to distinguish automated diagnostic detection related to EEG recording. The proposed methods used for signal detection and signal processing are Empirical Mode Decomposition (EMD) feature extraction and Artificial Neural Networks (ANNs) for classification. Mel-frequency cepstral coefficients (MFCCs) are used by the first-order differential method for distinguishing between normal and abnormal cases by reducing the amplitude factors that accelerate detection. Traditional epileptic diagnosis is based on rigorous tests performed by neuroscientists from long-term EEG recordings that require the presence of seizures. After pre-fixing the symptom signs such as mean, difference, deceit, kurtosis, and standard deviation. The classification accuracy obtained using this method is the neural transmission network (FFNN) and vector support machine (SVM). To classify associated EEG signals in the “normal” or “epileptic seizure” category based on the extracted features. This method achieves the highest accuracy of 99.44%, the sensitivity of 99.23% and the specificity of 100% respectively of the data set can be obtained using the proposed system.