DEVELOPMENT OF PATIENT REMOTE MONITORING SYSTEM FOR EPILEPSY CLASSIFICATION
Sunil Kumar Prabhakar, Harikumar Rajaguru
Bannari Amman Institute of Technology, India
Epilepsy is one of the most commonly occuring neurological disorders and is characterized by hypersynchronous neuronal firing. The unexpected and continuous electrical disturbances in the brain can be very disturbing to the patient thereby affecting the overall quality and happiness of the patient’s life. Electroencephalography (EEG) signals helps in the analysis and diagnosis of epilepsy by recording the activities in the cortical regions of the epileptic patient. Epilepsy is characterized by the abnormal EEG signal flow. The recordings of the EEG signals are too huge to process and hence the necessity of dimensionality reduction is mandatory. The total number of epileptic patients analyzed in this work is twenty. Initially the dimensions of the EEG data are reduced with the help of Fuzzy Mutual Information (FMI). The dimensionally reduced data is then transmitted with the help of a 2 x 2 Orthogonal Space Time Block Coded Multiple Input Multiple Output-Orthogonal Frequency Division Multiplexing (OSTBC MIMO-OFDM) system. As the OSTBC MIMO-OFDM System suffers a high Peak to Average Power Ratio (PAPR), a unique approach of Singular Value Decomposition Based Partial Transmit Scheme (SVD-PTS) is proposed to reduce the PAPR and Bit Error Rate (BER) at the receiver side. Also at the receiver side, Gaussian Kernel Based Support Vector Machine (G-SVM) with a Kernel value of 10 is utilized thoroughly in this work. The performance metrics such as Specificity, Sensitivity, Time Delay, Quality Value, Performance Index and Accuracy are analyzed here. Results show that an average accuracy of 95.38% is obtained, average perfect classification rate is 90.76%, time delay is 2.19 seconds and quality value is 20.53.