Purchase this article with an account.
P.S. Binder, L. Warden, A.W. Dreher; Neural Network to Predict Patient Preference for Wavefront Guided Spectacle Correction . Invest. Ophthalmol. Vis. Sci. 2006;47(13):1207.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To correlate the specific aberration type and magnitude with patient preference of a spectacle lens including (HOA) correction versus conventional lens.
67 patients with no previous history of refractive surgery or ocular abnormalities were enrolled from 7 clinical practices. Each patient underwent a manifest refraction by a clinician and a wavefront analysis (Z–ViewTM aberrometer, Ophthonix, Inc. San Diego). Two pairs of glasses were made, a conventional lens from the manifest (conventional, cv) and wavefront guided lens (iZonTM, wf) from the Z–View examination. After each pair was worn for one week, patients answered a questionnaire for their preferences. A value of 1 was assigned for wf preference and 0 for cv preference. A back–propagation neural network (NN) with one hidden layer using HOA as input parameters was made. The output was a number between 0 (cv preference) and 1 (wf preference).
As trefoil (Z3,1) increased from 0.02 D to 0.34 D, wf preference increased linearly, whereas patients preferred some spherical aberration (SA) (Z4,0) linearly between 0.07 and 0.13D but preference declined over 0.15D . In contrast, slight (<0.05 D) or more than 0.16D of Coma (Z3,3) was preferred to levels in between. When we combined the NN HOA into a single preference indicator, we found that 50% of the patients fit a profile that predicted preference for a given HOA corrective lens design.
We found non–linear relationships between the amount of an individual HOA and a patient’s preference for correcting it, i.e. some HOA are seen by patients as beneficial. A single metric was defined to suggest which patient (eye) might prefer a given combination of HOAs.
This PDF is available to Subscribers Only