Abstract
Purpose :
Cerebral visual impairment (CVI) resulting from brain injury is difficult to assess, since brain injury often also impairs the ability to communicate. Established approaches using preferential looking paradigms or visual evoked potentials are impracticable and hence generally not used clinically. We designed, built and validated a system that automatically determines visual thresholds in humans based on objective quantification of optokinetic responses.
Methods :
Our system uses a screen to present band-limited stimuli that move continuously at 12 deg/sec, while a desktop eye-tracker monitors eye movements. An automated algorithm determines whether subjects' eyes move smoothly with the stimulus, tolerating occasional reverse saccades. The output drives real-time feedback in the form of music, to keep subjects engaged in the task. It also drives adaptive adjustment of contrast or spatial frequency, to find the threshold at which smooth tracking is 75% successful.
Results :
First, we asked whether this system could make valid, reliable measurements of spatial vision. We used it to measure contrast thresholds at 8 different spatial frequencies in 4 healthy adults. The contrast sensitivity functions were (a) repeatable between subjects, (b) repeatable within-subject, and (c) inverted-U-shaped, suggesting that they reflected mechanisms of spatial vision.
Second, we asked whether the system could provide valid measures of acuity. In 8 healthy adults tested with and without optical correction, we found a good correlation (r=-0.88; p<0.001) between spatial-frequency tracking thresholds and LogMAR acuity measured using a tumbling-E chart. Furthermore, in 13 children with brain injury who could communicate, we found a good correlation (r=-0.74; p<0.01) between spatial-frequency thresholds and the LogMAR equivalent of their clinically assessed Snellen acuity.
Third, we asked whether the system was applicable to children with brain injury who could not communicate. 9 such children were tested. Their spatial-frequency thresholds spanned the same range as the 13 verbal children. We were able to perform repeated measurements with 15 of the 22 children, and the test-retest correlation was 0.86.
Conclusions :
These results indicate that our approach is a promising method of quantitative visual function assessment, and that it can be used even in cases where no other objective vision assessment is possible.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.