Open AccessIMPROVEMENT IN KERATOCONUS DIAGNOSIS USING MORPHO-GEOMETRIC VARIABLES WITH RNN NETWORKS
Author(s) -
R. Kanimozhi
Publication year - 2021
Publication title -
information technology in industry/information technology in industry
Language(s) - English
Resource type - Journals
eISSN - 2204-0595
pISSN - 2203-1731
DOI - 10.17762/itii.v9i1.83
Subject(s) - keratoconus , computer science , artificial neural network , artificial intelligence , sort , pattern recognition (psychology) , algorithm , cornea , medicine , ophthalmology , information retrieval
There is an eye disease called Keratoconus (KC) which has potential to cause visual acuity loss; hence, it can be considered as disability due to its severity. There are some limitations in current method in detecting cornea region’s boarder edge. Primary objective for the paper need to identify the structural description of disease’ asymmetry with the help of Morpho-geometric parameters relates with the keratoconous eyes along by means of slight visual control. It also includes the application of Recurrent Neural Network (RNN) analysis which is sort of Neural Network in which previous step’s output are sent to present step as an input. In order to determine most prominent correlation, Stepwise Discriminant Function Analysis is used in analyzing output. The Prominent correlation was identified between thinnest point in the anterior deviation and thinnest point in the posterior deviations of minor keratoconic cases. MATLAB R2014 software is used to implement the framework and analyses of simulation results were performed.