Automated Diabetic Retinopathy Prediction using Machine Learning Classification Algorithms
Main Article Content
Abstract
Diabetes is the most widely spread disease which takes place due to increase in blood sugar and when the body stops producing insulin. Diabetic Retinopathy (DR) is an ailment which has become a key cause of vision destruction in diabetic patients. It’s an eye disease which causes damage to retina. To protect a patient from complete blindness, prediction of DR in the early stage is necessary.
In this study, we focus on early prediction of diabetic retinopathy in an effective and affordable manner. We have considered Messidor image dataset for diabetic retinopathy which contains 1151 records. The ultimate aim of our research is to explore whether there is existence of diabetic retinopathy; by employing machine learning classification algorithms (Logistic Regression, K nearest neighbour, SVM, bagged trees) on the features that are extracted from outcome of various retinal images from the image dataset. As the data may contain outliers and noisy values, two types of data validation have been applied: cross validation and hold-out validation. For getting the best possible outcome, dimensionality reduction criteria using Principal Component Analysis has also been applied. In this research, the accuracy of logistic regression resulted as highest; giving 75.1% in cross validation and 82.6% in case of hold-out validation. Consequently, our findings imply that Logistic Regression is best suitable for DR prediction. Likewise, bagged trees have also turned up with 80% accuracy.
Downloads
Metrics
Article Details
Licensing
TURCOMAT publishes articles under the Creative Commons Attribution 4.0 International License (CC BY 4.0). This licensing allows for any use of the work, provided the original author(s) and source are credited, thereby facilitating the free exchange and use of research for the advancement of knowledge.
Detailed Licensing Terms
Attribution (BY): Users must give appropriate credit, provide a link to the license, and indicate if changes were made. Users may do so in any reasonable manner, but not in any way that suggests the licensor endorses them or their use.
No Additional Restrictions: Users may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.