Wavelet based approach for the investigation of vision impairments using single trial VEPs

Visually evoked potential (VEP) is being widely used for the investigation of vision abnormalities. The signals are recorded using non-invasive EEG electrodes from the Occipital Cortex while a subject is presented with a visual stimulus. By analyzing these responses, an ophthalmologist is able to determine the abnormalities in visual pathways of a person. The traditional method of analysis however, is centered on the detection of amplitude and latency values, in which long period of testing and averaging is required. This method could result in patients fatigue and affect the diagnosis accuracy. Hence, the wavelet based approach is investigated for the diagnosis of vision impairments. Biortogonal spline wavelet is used to decompose the VEPs and statistical features are extracted from the reconstructed signal for the analysis. k-Nearest Neighbor (kNN) classifier is employed for discrimination of vision impairments and the proposed method is able to produce 94.93% accuracy.