A metric-based image-formation model explains the improvement in subjective refraction using temporal defocus waves

Direct subjective refraction (DSR) is a novel method for refractive error measurements that uses temporal changes in defocus and a flicker minimization task. The computational models developed here are a framework for improving this clinical method.

This study aimed to model the measurement of refractive error with the DSR method, which uses rapid changes in optical power and a bichromatic (red/blue) stimulus.

The polychromatic point spread function of the eye was used to simulate the retinal image projected in DSR method, and an image quality (IQ) metric was defined based on the spatial frequencies of the retinal image. Three tasks were modeled: blur minimization (BM), monochromatic flicker minimization (MFM), and polychromatic flicker minimization or DSR. A metric was defined for each task and studied through focus in a ±3-D range. Whereas BM was modeled using only the IQ of the projected images, MFM and DSR metrics were a function of the IQ of the average retinal image and a metric to quantify the similarity (flicker) in the image. The width of the through-focus peak was used to compare between tasks, and different values of pupil size and spherical aberration were studied.

The through-focus 90% peak width was 0.48, 0.16, and 0.19 D for BM, MFM, and DSR tasks, respectively, which agreed well with previous experimental data. The 90% peak width increased for small pupils and with increasing values of spherical aberration in BM and MFM, but it remained relatively constant in DSR model.

The developed models explained previous experimental findings that reported a higher repeatability of the DSR compared with the traditional refraction method.