Automatic Measurements of Smooth Pursuit Eye Movements by Video-Oculography and Deep Learning-Based Object Detection

Abstract

AbstractPurposeThe purpose of this study was to develop a technique combining video oculography (VOG) with single shot multibox detector (SSD) to accurately and quantitatively examine eye movements.MethodsEleven healthy volunteers (21.3 ± 0.9 years) participated in this study. Eye movements were recorded while tracking a target using a custom-made eye tracker. The subjects were asked to fixate their focus on the nose of the rabbit-like target (visual angle was 0.1°), which was manually moved to a distance of 1 meter by the examiner during the eye movement test. The test produced 500 images from the VOG external camera and these images were divided into 3 groups (300, 100, and 100) for training, verification, and testing. The performance of the SSD was evaluated with 75% average precision (AP75), and the relationship between the location of the fixated target (calculated by the SSD) and the positions of both eyes (recorded by the VOG) was analyzed.ResultsThe AP75 of the SSD on one class of targets was 97.7%. The horizontal and vertical target locations significantly and positively correlated with the horizontal and vertical both eye positions (adjusted R2 ≥ 0.955, P < 0.001).ConclusionsOur findings suggest that VOG with SSD is suitable for the evaluation of eye version movements in standard clinical assessments.Translational RelevanceThe combination of VOG and SSD can be used to evaluate the SPEM, and this method can be translated into clinical settings without changing the testing methods.