Characterizing selected visual anomaly through wavelet decomposition of evoked responses

Abstract

Evoked responses provide valuable information about the reaction of the brain towards external stimuli. These reactions are observed through the electroencephalogram (EEG) recordings using non-invasive EEG surface electrodes. For this research, four checkerboard patterned visual stimuli were used to induce brain responses related to the visual system. The responses were later recorded and analyzed using wavelet decomposition for characterizing selected visual anomaly, namely myopia. Although the analysis of visually evoked responses (VEP) for visual anomaly can be traced back to the early 80's, the advancement of digital signal processing techniques has yet to be fully utilized for investigating these responses. Therefore, this work serves as an important milestone for investigating the characteristics of VEP's using biortogonal spline wavelet to define the time-frequency characteristics of the signals. Probablistic neural network (PNN) was used to evaluate the performance of the extracted features in discriminating the myopia as well as healthy controls and the proposed method is able to achieve a maximum accuracy of 95.07%.