Retinal Image Analysis for ROP Plus Diagnosis and Detection

Abstract

In this work an attempt is made to address the Retinopathy of prematurity (ROP) plus disease analysis and verification of the developed methods and algorithms in consultation with the ophthalmology expert.  Few of the test cases were used in analysing the vessel width dilation and tortuosity of the retinal image features for setting and confirming the thresholds to classify the cases as normal and abnormal.

The adopted thresholds are testified for confirmation by running the algorithmic simulation for a set of selected test cases chosen from the study reported by other groups. The results of which showed the set threshold as good measures in predicting. Further to gauge the accuracy of the algorithm expert opinion were collected and compared with the results of algorithmic simulation for a selected test cases. The outcome of the approach showed the accuracy of the algorithm to be >75%.