Please use this identifier to cite or link to this item:
Title: Cálculo de la potencia de una lente intraocular mediante técnicas de machine learning
Author: Carmona González, David
Tutor: Alférez Baquero, Edwin Santiago
Others: Universitat Oberta de Catalunya
Keywords: biometrics
machine learning
intraocular lens
Issue Date: 5-Jun-2018
Publisher: Universitat Oberta de Catalunya
Abstract: The purpose of this master's thesis projecte is to develop a predictive algorithm to obtain the power of an intraocular lens to be implanted in an eye. Retrospectively, a total of 260 eyes biometric characterized with intraocular lens implanted and their corresponding refractive outcomes. Were used to get the best refractive outcomes, several regression algorithms were implemented to predict the dioptric power of the lens that will be implanted in a surgery. The best result was a stacking regression model with a R2 equal to 0.9863 and a RMSE equal to 0.6102. Comparing with other intraocular lens power calculation formulas, the outcomes were MAE of 0.34 ± 0.26 D (0.00/1.07) and the percentage of eyes within ± 0.50 D were 75%. A non-parametric regressive model with machine learning techniques is implemented, capable of competing with the best intraocular lens power calculation formulas.
Language: Spanish
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
dcarmonagonTFM0618memoria.pdf1.26 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons