Prediction Of Anterior Chamber Diameter Based On Artificial Neural Networks

The main technology for eye vision remedial are excimer Laser surgery and intraocular lens implantation. According to the location of surgical implantation, intraocular lens implantation can be divided into anterior chamber intraocular lens implantation and rear intraocular lens implantation.In anterior chamber intraocular lens implantation, correctly forecasting the anterior chamber diameter plays an important factor in the success of the surgery .Formulas existing for anterior chamber diameter prediction consider less parameters and have low accuracy ,so they cann't meet the need of AC-IOL surgery. In this paper, the use of statistical analysis and artificial neural network approach to predict the anterior chamber diameter is introduced. Forecast parameters includs axial length of eyes, the maximum of corneal curvature KD, the minimum corneal curvature KF, 3mm radius of curvature of the cornea of the topography and the depth of anterior chamber .This paper conducted a multivariate regression analysis, followed by a stepwise regression analysis to omissions the parameters that have no significant effect on anterior chamber diameter prediction. In BP network model, this paper introduces method that uses the adaptive learning rate and momentum to overcome the weakness of slow convergence and easy to be trapped into local minima in the BP network learning process. In BP network model,this paper introduces a mothod that uses the most neighboring clustering algorithm to determine the center of the radial basis function networks.This paper mainly uses the MATLAB to realize the prediction of anterior chamber diameter of the regression analysis model,the BP network model and the RBF network model separately. The experiment shows that the artificial neural network model is more accuracy than linear regression analysis model ,meets the need of the AC-IOL surgery and can be used to the prediction of the anterior chamber diameter.