Comparison of retinal image evaluation techniques in novice clinicians.

Abstract

Retinal fundus evaluation is learned through experience and training. This study aimed to determine the image presentation characteristics and the accompanying evaluation techniques, which led to the most accurate and efficient retinal pathology detection method. Phase I included 25 novice clinicians asked to evaluate 14 different pathologies using spatial versus temporal image presentations. Phase II included 25 different novice clinicians asked to evaluate five different simulated pathologies at three different pixel sizes presented in both spatial and temporal image presentations. Accuracy and speed of recognition were evaluated between the spatial and temporal presentations of the same simulated pathology. In phase l, subjects were significantly faster at simulated pathology detection using a temporal presentation with a 95% accuracy rate versus a spatial presentation with a 79% accuracy rate. In phase II, subjects demonstrated significant differences in speed of detection using the temporal technique at all 3 pixel number sizes with the greatest difference in detection times shown at the smallest retinal defects. Accuracy and speed of recognition in simulated pathology assessment were improved in a temporal presentation and the greatest improvements were demonstrated at the smallest pixel numbers.