Saccadic vector optokinetic perimetry : a technique and system for automated static perimetry in children using eye tracking

Abstract

Perimetry is essential to identify visual field defects in disorders of the visual pathways. In compliant adults, automated static perimetry (ASP) is the preferred method of visual field assessment. However, children under 10 years have difficulty with the visuo-motor task and constant fixation required. Manual kinetic perimetry is often used for children as it can be adapted to a child’s age. However, it suffers from many of the problems inherent to ASP. In infants perimetry is limited to the “confrontation” technique which can be imprecise and does not generate quantitative data. The lack of reliable ASP in children and quantitative perimetry in infants is a longstanding clinical problem. The aims of this research were to (i) develop, and (ii) clinically evaluate, a technique for ASP in children which utilises “eye tracking”. The first part of this research was concerned with the development of the technique, termed “Saccadic Vector Optokinetic Perimetry” (SVOP). The system comprises a personal computer, display screen, and an X50 eye tracker (Tobii Technology, Sweden). The eye tracker is noncontact and provides data on (i) eye position in 3D space, and (ii) the point of gaze. This allows the screen position of “test stimuli” to be calculated, and eye gaze responses to the “test stimuli” to be assessed in “real time”. A software algorithm was developed to determine if “test stimuli” have been perceived based on the direction, amplitude and latency of a subject’s gaze response. A feasibility study was conducted with 29 subjects comprising 4 groups: (i) healthy adults, (ii) healthy children, (iii) adult patients with visual field defects, and (iv) child patients with visual field defects. Subjects performed SVOP tests which replicated the Humphrey Field Analyser (HFA) C-40 screening test with a stimulus size of Goldmann III and intensity of 14dB. Subjects able to do so also performed equivalent HFA C-40 tests for comparison. In healthy subjects 99.1% of SVOP test points were in agreement with a healthy visual field. In patients with visual field defects, 89.8% of test points were in agreement with HFA equivalent tests. The visual field defects identified using SVOP in the child patients were consistent with their clinical findings. A clinical evaluation of SVOP was undertaken in the second stage of this research with 122 subjects comprising the same four subject groups as in the feasibility trial. An “ideal” test protocol resulted in 8 uniocular visual field tests for each subject comprising 4 SVOP tests and 4 HFA tests. In children where uniocular testing was not tolerable, two binocular SVOP tests were performed. The sensitivity and specificity of the SVOP tests were computed using a direct comparison with reliable HFA tests, and repeatability of SVOP and HFA tests were assessed using Cohen’s kappa coefficient. In child patients unable to provide a reliable HFA test, their clinical history, other clinical findings and the repeatability of their SVOP tests were used to assess the SVOP results. The overall sensitivity and specificity of the SVOP testing was 72.7% and 96.8% respectively. The sensitivity had a greater variation than the specificity amongst the different subject groups. The repeatability of SVOP tests was slightly reduced as compared to the HFA tests across all groups with kappa coefficient’s of 0.65 and 0.74 for SVOP and HFA respectively. In child patients without reliable HFA equivalent tests the SVOP results could commonly be associated with other clinical findings and repeatable testing added to the confidence in the reliability of these cases. The developed SVOP technique performs well with accurate eye tracking data and an attentive child. It has proved extremely useful in identifying and monitoring visual field defects in several child patients who required regular visual field assessment.