Detecting Retinal Detachment using images with Machine Learning

This research is on how AI and machine learning can be used to detect retinal detachment using images.

Retinal detachment occurs when your retina (a light-sensitive layer of tissue in the back of your eye) is pulled away from its normal position at the back of your eye. Symptoms include increased eye floaters, sudden flashes of light in vision, and dark shadows in the middle or on the sides of vision. It can eventually lead to blindness. Therefore, it is important that we detect it early on and treat it. This research is on how AI and machine learning can be used to detect retinal detachment using images. Before building my models, I did data pre-processing on retinal images data from four different hospitals. Then I built several different machine learning models, including Logistic Regression, Ridge Classifier, Decision Tree Classifier, K-nearest neighbor, Random Forest Classifier, SVM, MLP Classifier, Neural Networks, and Transfer Learning. I then assessed their accuracy using certain metrics. After building the models, I calculated the accuracy score, precision score, recall score, F1 score, and built Confusion Matrices for all of the models. Our results showed a very high accuracy in all of them; they were all in the 90% to 99% range. Therefore, I can assume that Retinal Detachment can be diagnosed using Machine Learning methods. In summary, retinal detachment is a serious issue and needs to be detected and treated. Using our machine learning models, I achieved great results.

Explore More!

Research Paper