Back to Freepaper Session

Development of an algorithm to predict vault (distance between posterior chamber phakic IOL and crystalline lens) in patients implanted with a posterior chamber phakic refractive IOL (Visian ICL)

Session Details

Session Title: Phakic IOL Implantation II

Session Date/Time: Wednesday 09/10/2013 | 08:00-10:30

Paper Time: 09:03

Venue: E104-105 (First Floor)

First Author: : A.Villarrubia SPAIN

Co Author(s): :    A. Cano   C. Garcķa-Alonso   A. Gomera        

Abstract Details

Purpose:

Distance between the posterior surface of a Visian ICL phakic IOL (Staar, Switzerland) and the anterior capsule of the lens (vault) has been advocated as the main reason for the development of anterior opacities in patients with myopia implanted with this phakic IOL: the smaller the vault, the higher the rate of anterior cataract. Moreover, vault between 300 and 750 microns, usually gives us the idea that the Visian ICL is properly situated in the sulcus; and this means that the total length of the ICL match with the sulcus-sulcus distance in a safe manner. We have developed an algorithm to predict vault in almost 100% of the patients implanted with a Visian ICL

Setting:

Instituto de Oftalmologķa La Arruzafa de Córdoba (Spain)

Methods:

During 2011 and 2012, we have implanted 170 Visian ICL, but we have only included in this study the cases of myopia in which we did not change the recommendation of Staar (online calculator): this means 125 eyes of 64 patients. In all patients we have measured vault with OCT-Visante (Carl Zeiss, Germany) one month postoperatively. We included patients in 3 different groups of vault: group A (0 - 299 microns); group B (300 – 750 microns); group C (751 or more microns). These 3 groups of vault have been related with other variables such age, spherical and cylindrical refraction, keratometry, anterior chamber depth (ACD), corneal thickness (CT), white to white distance, steep and flat axis and length of the Visian ICL. The relationships between groups and these other variables have been made using statistical analysis like C4.5 (weka diagram) and Kruskal-Wallis method

Results:

CT, spherical and cylindrical refraction and the length of the ICL selected are relevant to explain the model. ACD and keratometry do not influence this decision tree. Total Number of Instances is 125. Correctly Classified Instances are 88,8%, Incorrectly Classified Instances are 11,2%, Kappa statistic is 0,7567, Mean absolute error is 0,1232, Root mean squared error is 0,2482, Relative absolute error is 36,5884% and Root relative squared error is 60,7164%. The algorithm obtained is so complex (29 arms for a total number of 125 instances). However, with this analysis we are able to obtain an impressive level of classification in the group B (100%), a moderate level of classification in the group A (73,33%) and an acceptable level of classification in the group C (61,53%)

Conclusions:

With complex statistical analysis, it is possible to achieve an algorithm to predict the vault in a great number of patients after implantation of a posterior chamber phakic refractive IOL (Visian ICL). In cases in which and extreme and unsafe vault is predicted, a modification in the selection of the selected length of the Visian ICL could be of interest to obtain a safe vault in almost 100% of the patients

Financial Interest:

NONE