To establish the repeatability and reproducibility of a virtual refraction
process using simulated retinal images
With simulation software, aberrated images corresponding with each step of the refraction
process were calculated following the typical protocol of conventional subjective refraction
. Fifty external examiners judged simulated retinal images
until the best sphero-cylindrical refraction
and the best visual acuity were achieved starting from the aberrometry
data of three patients. Data analyses were performed to assess repeatability and reproducibility of the virtual refraction
as a function of pupil size and aberrometric profile of different patients.
SD values achieved in three components of refraction
, and J45
) are lower than 0.25D in repeatability analysis. Regarding reproducibility, we found SD values lower than 0.25D in the most cases. When the results of virtual refraction
with different pupil diameters (4 and 6 mm) were compared, the mean of differences (MoD) obtained were not clinically significant (less than 0.25D). Only one of the aberrometry
profiles with high uncorrected astigmatism shows poor results for the M component in reproducibility and pupil size dependence analysis. In all cases, vision achieved was better than 0 logMAR. A comparison between the compensation obtained with virtual and conventional subjective refraction
was made as an example of this application, showing good quality retinal images
in both processes.
The present study shows that virtual refraction
has similar levels of precision as conventional subjective refraction
. Moreover, virtual refraction
has also shown that when high low order astigmatism is present, the refraction
result is less precise and highly dependent on pupil size.