Analysis of vertical eye movements in Parkinson’s disease and its potential for diagnosis

Abstract

It is well known that eye movements are highly affected by Parkinson’s disease. The majority of studies related to effects of Parkinson’s disease on eye movements have been performed for rapid eye movements during sleep. However, in the current study, eye movements during resting-state (eyes-open and eyes-closed conditions) were studied to evaluate Parkinson’s disease. To measure eye movements, vertical electrooculography (VEOG) was used. Using extracted features in time-, frequency-, and time-frequency domains of VEOG time-series, a classification analysis between healthy subjects and Parkinson’s disease patients in OFF- and ON-medication states was performed. The most-informative features for an error-correcting output codes support vector machine classifier were selected according to the multiple-comparison corrected p-values. VEOG data obtained 69.10 % and 87.27 % discrimination accuracy for OFF- and ON-medication states, respectively. Interestingly, higher discrimination was obtained for the lower frequency contents of VEOG time-series (0.1–1.25 Hz). The most discriminative features were related to the variation of amplitude and frequency content for OFF-medication, while in ON-medication the features according to the variation of VEOG amplitudes were removed from the most discriminative features list. Based on the obtained results, it was concluded that vertical eye movements of Parkinson’s disease patients had lower amplitude variation compared with healthy subjects in OFF-medication, while levodopa prescription increased such variation in vertical eye movements during the eyes-closed condition and decreased the variation during the eyes-open condition. Levodopa prescription possibly affects the amplitude variation of VEOG time-series, while it had no effect on the movement rates (frequency contents) of vertical eye movements.