*Corresponding author:
Ali S Saad, Department of Biomedical Technology, College of Applied Medical Sciences, King Saud University, P. O. Box 10219, Riyadh 11433, Saudi ArabiaReceived: August 30, 2018; Published: September 06, 2018
DOI: 10.26717/BJSTR.2018.08.001702
To view the Full Article Peer-reviewed Article PDF
Patients with long term dry eye (DE) could suffer cornea damage. To date, there is no universal diagnostic test to detect eye dryness. Recently, an automated diagnostic method of DE using thermographic images provided a good binary classification between normal and DE subjects. The present work focuses on the automatic diagnosis of DE patients using tear ferning (TF) images. The TF images for 100 subjects were used to classify their degree of dryness automatically (without human interaction) using a five-point grading scale. First, from each TF image a vector characteristic (VC) was computed to represent each patient during the automatic classification. Then, it was compared to the VC references representing the five points grading scale. Next, each subject’s TF image was assigned to the closest reference. The classification of each patient’s degree of dryness using the vector characteristic is based on a decision tree approach using “See5”: tools for data mining that generates decision trees for classification purpose. The fully automatic classification (diagnosis) provides a promising result compared to the manual classification by two experts. It was found that 81% of the automatic classification perfectly matched the manual ones. 12% of the automatic grades were misclassified to the next grade (i.e. one grade difference between the automatic and manual grading). Only 7% were completely misclassified. Future work on the VC and the algorithm of classification using synthetic TF images could further improve the automatic classification of dry eye patients.
Abstract | Introduction | Methods and Materials | Results | Discussion | Conclusion | Acknowledgement | References |