Purchase this article with an account.
S Schutte, SP W van den Bedem, F van Keulen, HJ Simonsz; Non Deterministic Optimization to account for Anatomical and Physiological Variability in Strabismus Surgery Modeling . Invest. Ophthalmol. Vis. Sci. 2002;43(13):221.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: Current computerized strabismus models do not account for the variability of anatomical and physiological parameters that constitute the model, instead they employ averages. To account for this variability the bounded-but-unknown-uncertainties optimization algorithm was applied to the Simonsz-Robinson computer model. Methods: All parameters influencing the results of surgery, like binocular vision, were categorized. This study, however, was limited to anatomical and physiological parameters as used in the model. In a preanalysis the influence of these parameters on the prediction of the model was determined, in a simulation of infantile esotropia and of a superior oblique palsy. The quality of the simulated postoperative result was rated by the angles of strabismus in primary position and, to a lesser extent, in secondary positions. The model was adapted for Linux and connected to the optimization software. Optimization was carried out by varying the amount of surgery and evaluating the postoperative result. For each step, combinations of the most prominent parameters were examined to find the set of values that provided the poorest postoperative result. Accordingly the optimization resulted in the best advise for the worst possible combination of parameters. Results: Preanalysis showed that the spring constant in Passive Rotation (PR) of the eye (muscles detached) was the most influential parameter. The spring constant of the Innervated Muscles (IM) was second, the relative spring constant of vertical rectus and oblique eye muscles third. The influence of all other parameters could be neglected. For horizontal strabismus the optimized model advised surgery on two horizontal muscles in approximately equal amounts. Optimization with uncertainties showed that for PR in the order of 0.1 [g/°] the influence of IM is minimized (Fig.) View OriginalDownload SlideView OriginalDownload Slide Conclusion: Application of optimization with bounded-but-unknown-uncertainties is currently hampered by the lack of information on the spring constant in passive rotation of the eye, which unfortunately, has the most influence on the prediction.
This PDF is available to Subscribers Only