Clinical evaluation of an automated subjective refraction method implemented in a computer-controlled motorized...

To investigate a new algorithm to perform an automated non-cycloplegic refraction in adults.

Fifty healthy subjects were measured twice (test-retest) with the new automated subjective refraction method and with the conventional clinician subjective refraction procedure. Objective refraction was also measured with the Grand Seiko WAM-5500 autorefractor. The new automated method was inspired on the root finding bisection algorithm and on the Euclidean distances in the power vector domain. The algorithm was implemented in a computer that was synchronized with a customized motorized phoropter. Repeatability was mainly assessed with the within-subject standard deviation (Sw) and accuracy was mainly assessed with the limits of agreement.

The within-subject standard deviations of the power vector components (M, J0, J45) obtained for the right eye are (±0.13, ±0.04, ±0.05)D and (±0.17, ±0.03, ±0.07)D, respectively, for the clinical and the automated subjective refraction methods. The limits of agreement (with the clinical method) for the automated and the objective methods are, respectively (±0.56, ±0.18, ±0.31)D and (±0.77, ±0.15, ±0.18)D. Similar results are obtained for the left eye.

The proposed automated method is repeatable and more accurate than objective techniques in healthy adults. However, it is not accurate enough to replace the clinical subjective refraction yet and it should be tested in a wider population in terms of age, refraction and different ocular conditions. Despite these important limitations, this method has been shown to be a potentially valuable method to improve the access to primary eye care services in developing countries.