KeratoEL: Detection of keratoconus using corneal parameters with ensemble learning

Area of Science:

• Ophthalmology
• Medical Technology
• Artificial Intelligence

Background:

• Keratoconus is a progressive eye condition causing corneal thinning and bulging, leading to distorted vision.
• Early detection of keratoconus is challenging due to its subtle initial presentation.
• The condition necessitates accurate diagnostic tools for timely intervention and management.

Purpose of the Study:

• To develop and evaluate an ensemble-based machine learning model for early keratoconus detection.
• To assess the efficacy of the KeratoEL model in identifying keratoconus compared to existing methods.
• To determine the optimal number of features for accurate keratoconus classification.

Main Methods:

• An ensemble machine learning (ML) technique, KeratoEL, was proposed, integrating Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), and Artificial Neural Network (ANN).
• Data preprocessing involved feature elimination, followed by feature importance analysis using Extra Trees Classifier.
• The top 45, 30, and 15 features were selected as input datasets for model training and evaluation.

Main Results:

• The KeratoEL model achieved high accuracy rates of 98.0%, 98.9%, and 99.8% with 45, 30, and 15 features, respectively.
• Experimental results demonstrated that the proposed ensemble model significantly outperformed existing machine learning models.
• The model's performance indicates its robustness in classifying keratoconus cases.

Conclusions:

• The KeratoEL model effectively detects keratoconus at an early stage by leveraging an ensemble of ML algorithms (SVM, DT, RF, ANN).
• The proposed approach shows superior performance compared to existing methods, highlighting its potential for clinical application.
• This study underscores the value of ensemble ML in enhancing early detection and management strategies for keratoconus.