Abstract

Study Design:Cross sectional screening method to detect catarct with the help of Artificial Inteligence (AI)
Method: Captured images’ texture features: uniformity, mean intensity, entropy, contrast, energy and standard deviation; are first computed, populated and mapped with the diagnostic opinion by ophthalmologist. Accuracy is added by comparative study and development of machine learning algorithms to assign weights to multiple texture parameters and determine thresholds to detect and grade cataract based on the degree of its maturity. A mathematical model is proposed to determine weights for texture parameters. Grading is accomplished by applying the K-means clustering algorithm on collected parameter data post assigning suitable weights to them.
Result:The sensitivity, specificity and accuracy of the system are 95.8%, 96.4%, and 96%, respectively, which proves its clinical efficacy.
Conclusion: Our study proposed that AI based cataract’s screening can be done with good accurecy.