An Automatic Ametropia Examination based on EEG Signals

Abstract

Abstract Regular ametropia examination of human eye can effectively prevent the aggravation of ametropia. However, the use of the traditional ametropia examination method is limited to complicated manual operations and necessary communication between doctors and patients. In this study, we proposed a novel ametropia examination method, which determine the degree of ametropia by automatically processing and analyzing the event-related potentials (ERP), one kind of evoked electroencephalogram (EEG) signal. Specifically, we studied the correlation between one’s ametropia degree and his ERP signal features, and established the ametropia classification model based on ERP for the first time. In order to stably induce ERP from subjects, we designed a compound visual stimulus paradigm, which integrated three types of stimulus pictures to provide non-single stimulus. Furthermore, to realize automatic signal processing, we devised the eye movement artifact removal algorithm based on standard deviation as the threshold, the ERP component location algorithm based on peak and valley, and feature extraction formulas. In the test experiment, the examination accuracy of the method for low, moderate, and high ametropia is 86.29%, 80.95% and 90.98% respectively. And the average time of ERP feature extraction is 0.012 seconds. When 1.5 standard deviation was set as the threshold, the method can effectively identify and eliminate eye movement artifacts, and retain effective EEG signals. Compared with the traditional ametropia examination method, a novel ametropia classification model based on ERP proposed in this study can realize automatic ametropia examination by oneself and is more suitable for daily family examination.