Predicting patient-reported quality of vision through multifocal optics

To explore the agreement between computationally modelled point-spread functions (PSFs) and multifocal soft contact lens wearers' descriptions of image quality.

In Phase 1, participants were fitted with a single vision soft contact lens (SCL) in the left eye, and four different multifocal (MF) SCLs in the right eye in randomised order. Aberrometry measures of the MFSCL-wearing right eye were collected while participants viewed a point source monocularly using their left eye and drew their perception of the source. The predicted right eyes' PSFs were computed from the aberrometry data using a geometrical optics PSF calculation. In Phase 2, participants completed a Likert item questionnaire designed to identify salient features in the PSFs. Consensus among the participants about these features was assessed quantitatively using Yule's Q. Participants also rated the similarity between pairs of PSFs. Non-metric multidimensional scaling (MDS) and Procrustes transformation analyses were used to derive and compare image similarity spaces for the drawn and computed PSFs, respectively. Lastly, two observers completed an image matching task, in which they selected the best-matching drawn PSF for each computed PSF.

Significant similarities between computed and hand-drawn PSFs were demonstrated both qualitatively and quantitatively. Inter-rater agreement for image features was statistically significant (Yule's Q = 0.63; p < 0.05). Stress values for three-dimensional MDS configurations were <0.1, indicating excellent correspondence between MDS configurations and observed similarity ratings. Procrustes analysis confirmed strong concordance between hand-drawn and computed 3D configurations (congruence coefficient = 0.90). Image matches were inconsistent with random guessing (p < 0.001).

Results indicate good agreement between the PSF image quality drawn by participants and that predicted from aberrometry measures. This study has demonstrated potential for leveraging the correspondence between subjectively perceived and computationally derived representations of image quality in both clinical and contact lens research and development processes.