Deep Learning-Based Automated Detection and Classification of Diabetic Retinopathy Using MobileNetV2 and DenseNet201

Govinda Ravindra Kumar, Aswin Kumar (2024) Deep Learning-Based Automated Detection and Classification of Diabetic Retinopathy Using MobileNetV2 and DenseNet201. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (2MB) | Preview
Preview	PDF (Configuration Manual) Download (6MB) | Preview

Abstract

Diabetic Retinopathy (DR) is one of the most common causes of blindness in global health today, early screening is important in preventing the diseases progression and complications. The existing approaches in diagnosing TB are time-consuming, involves the use of expert personnel which is barely available for timely diagnostic intervention. Although the automated solutions based on deep learning have been demonstrated as feasible, a range of unvarying and efficient classification across all the levels of DR stage has not been achieved. This study leverages state-of-the-art deep learning models, MobileNetV2 and DenseNet201, to classify retinal images from the Kaggle Retinopathy dataset into five DR stages: It can be None, Mild, Moderate, Severe and Proliferative Diabetic Retinopathy abbreviated as No DR, Mild, Moderate, Severe, Proliferative DR. Normalization was used with the aims to make the model more launches and insensitive to the input data values, as well as, data augmentation was performed to minimize class over/under-representation. Fine-tuning these models was done while employing common metrics, including accuracy rate, precision, recall rate, and F1-coefficient. The results therefore validate these models, with MobileNetV2 attaining 93.4% accuracy and F1-score, while DenseNet201 attained 92.5% accuracy and F1-score. In view of these, it is evidenced that combining efficient deep learning techniques for scalable, accurate, and efficient DR classification allows enhanced diagnostic reliability in clinical practice.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Zahoor, Sheresh UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science R Medicine > RE Ophthalmology R Medicine > Healthcare Industry
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	19 Jun 2025 15:55
Last Modified:	19 Jun 2025 15:55
URI:	https://norma.ncirl.ie/id/eprint/7950

Actions (login required)

View Item