ITDS | SPMS

INNOVATIVE DIABETES PREDICTION USING RETINA FUNDUS IMAGES: A SMARTPHONE APPROACH

Project Description
Supervisor

Diabetes, a chronic condition affecting millions worldwide, can lead to severe complications if not detected early. This study focuses on developing an advanced Convolutional Neural Network (CNN) model to predict diabetes by analyzing retinal fundus images. Recognizing the limitations of existing models, we created a comprehensive local dataset to capture diverse diabetic retinopathy manifestations. We trained and evaluated four deep learning algorithms: a custom CNN, Inception-V4, Visual Geometry Group VGG-16, and DenseNet on this retinal fundus image dataset. The CNN model emerged as the top performer, achieving an accuracy of 96.35%, precision of 98.53%, and recall of 96.22%, with a strong F1-Score of 92.98% despite a relatively high loss of 28.66%. Inception-V4 followed with 91.28% accuracy and the highest ROC-AUC score of 97.24%, though its precision was lower at 82.48%. The VGG-16 and DenseNet models showed lower overall accuracy (77.57% and 78%, respectively), but DenseNet demonstrated the lowest loss (10.28%) and a higher F1-Score (79.88%) compared to VGG-16. The best-performing model was deployed in a user-friendly smartphone application that captures and processes retinal images, offering real-time diagnostic results. This innovation aims to enhance early detection, reduce complications, and improve health outcomes, particularly in underserved and remote areas. Future research directions include improving the models' predictive ability, applying advanced routing algorithms in capsule network models, comparing with machine learning models, and incorporating real-time diabetes detection via a detachable miniature ophthalmoscope.

Artificial Intelligence: Artificial Intelligence (AI) research in Computer Science and Information Technology explores the development of systems and algorithms that can perform tasks typically requiring human intelligence. These tasks include problem-solving, decision-making, learning, perception, and language understanding. AI research is a vast and interdisciplinary field, encompassing a variety of areas such as machine learning, natural language processing, computer vision, robotics, and cognitive computing. Machine learning (ML) is a core component of AI research. It focuses on creating algorithms that allow computers to learn from data, improve over time, and make predictions or decisions without explicit programming. Subfields of ML include supervised learning, unsupervised learning, reinforcement learning, and deep learning, with applications ranging from image recognition to recommendation systems. Natural language processing (NLP) is another critical area within AI research. It aims to enable machines to understand, interpret, and respond to human language. Researchers in NLP work on tasks like speech recognition, language translation, and sentiment analysis, making it possible for computers to interact with humans in a more natural and intuitive way.