EXTRACTION OF VEP SIGNAL FEATURES BY USING QUADRATIC BIORTHOGONAL B-SPLINE WAVELET

• Main Author: Ahmad, Maula
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36250
• Institution: Institut Teknologi Bandung

Eyes are important senses for humans. Despite its vital role, this organ is susceptible to disease. One of the disturbances that can occur is Optic Neuritis, in which the process of demyelination occurs. To detect this disease early, one method of testing is Visual Evoked Potential (VEP). VEP is a clinical testing method that is performed to determine brain response to visual stimulus through EEG signal reading in the occupital lobe Oz. From these signals, the latency values of P75, P100 and P145 will be obtained to assess visual conditions. This research was conducted to optimize the VEP signal feature extraction process using quadratic biorthogonal b-splines wavelet. To optimize the feature extraction process, 5 different sampling rates were chosen, namely 256, 224, 192, and 160 for the data retrieval process. The stages in this study are the EEG circuit calibration process that will be used for data acquisition using input from EEG simulator with an input frequency of 5Hz and a peak amplitude of 30uV. Furthermore, the data acquisition process with 5 different sampling rates were taken in 13 male subject and 10 female subject. In the feature extraction stage, the data will be filtered using a bandpass filter with a frequency band of 1-50Hz to reduce noise. Then the data will be decomposed using wavelet transforms and VEP signal will be reconstructed by using D3 and D4 detail levels. From the results of the reconstruction, P75, P100 and P145 features will be obtained. The data analysis phase is done by comparing the standard deviation of the average results from the VEP feature extraction categorized based on the sampling rate. From the results of the study, the smallest standard deviation is 8.93 at 256 sampling rate, and the largest standard deviation value is 13.35 at 160 sampling rate.