Purchase this article with an account.
Guang-ming George Dai, Dimitri Chernyak; Algorithms for Reducing Post-Operative Induction of Spherical Aberration for LASIK Surgeries. Invest. Ophthalmol. Vis. Sci. 2017;58(8):5270.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To investigate treatment algorithms for reducing post-operative induction of spherical aberration for LASIK surgeries.
Following a previous approach [A. Fabrikant, G.-m. Dai, and D. Chernyak, “Optimization of linear filtering model to predict post-LASIK corneal smoothing based on training data sets,” Applied Mathematics 4, 1694-1701 (2013).] for using an optimized linear filter (OLF) model to reduce post-operative induction of spherical aberration for myopic LASIK surgeries, we proposed an improved approach that imposes a criterion such that the kernel does not have a wing that would cause instability during optimization. The optimization also includes both myopic and hyperopic data to increase the predictability of intended versus achieved refractions. In addition, various weighting factors are introduced during the optimization to fine-tune the optimized outcomes. A new technique is devised to find the best SNR (signal to noise ratio) value in the deconvolution. Finally, the power spectrums for the OLF as well as the inverse kernel are plotted against those of the previous kernels to make sure their similarity in the spatial frequency domain.
Analysis of the clinical study outcome [G. Dai, D. Chernyak, S. Kasthurirangan, and J. Tarrant, “Outcomes for clinical studies to reduce post-operative induction of spherical aberration for myopic LASIK surgeries,” ARVO abstract, 2016.] using the previous kernel indicated that it is too weak, resulting in sub-optimal reduction of spherical aberration in the clinical study. It was found that unbound kernels (kernels with wings) have the following problems: (1) no biological support; (2) unstable optimization; (3) no natural limit on the kernel size; (4) with fluctuating power spectrum. Hence, the bound kernels are introduced. Analytical solutions are obtained for the cutoff of bound kernels at zero first-derivative for 2-, 3-, and 4-parameter kernels. A multi-level optimization approach has been developed to enable the realization of attaining the optimized kernel both in speed and precision. The proposed new kernel results in similar ablation depth to CustomVue for myopia and about 25% deeper for hyperopia.
It is possible to further improve the predicted clinical outcome with a new optimized linear filter for reducing the induction of spherical aberration for LASIK surgeries.
This is an abstract that was submitted for the 2017 ARVO Annual Meeting, held in Baltimore, MD, May 7-11, 2017.
This PDF is available to Subscribers Only