International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025
p-ISSN: 2395-0072
www.irjet.net
Early Diseases Prediction via Retina Scan Using Deep Learning Miss.Akshata Dunagi1, Mr. Madivalappa Sajjan2 1Teaching Assistant, Department of Computer Science, Rani Channamma University, Dr.P.G.Halakatti Post
Graduate Center“Vachana Sangam”, Vijayapura, Karnataka India
2PG Scholar, Department of Computer Science, Rani Channamma University, Dr. P.G.Halakatti Post Graduate
Center “Vachana Sangam”, Vijayapura, Karnataka India.
---------------------------------------------------------------------***--------------------------------------------------------------------areas. The need for a automated, fast and accurate Abstract - The Ophthalmic disease like Glaucoma, diabetic classification system to support early diagnosis is need of the Retinopathy (DR) and Additional Macular Degeneration hour in present scenario. This project addresses the gap by Associated with age (AMD) are the major contributors to the developing the AI model that can classify the retinal images global vision loss. Their asymptomatic nature in early stages based on diseases and provides a non –invasive, scalable demands an automated diagnostic tool which has capabilities solutions for mass screenings and clinical support. of detection of diseases in early stages and its treatment. This project employs deep learning specifically convolution neural 3. OBJECTIVES networks (CNNs) and transfer learning processes, to develop a model that can classify the retina scan images into healthy or diseases categories. Using pertained models with fine tuning To develop a reliable AI-based system that detects common and image augmentations techniques the model achieves high eye diseases from retina scan images. classification accuracy. This work is the promising solution for To utilize advanced architectures of CNN along with transfer application in real world clinical applications and learning to enhance prediction accuracy. telemedicine solutions, improving access to early eye disease detection particularly in under resourced health care settings. To implement effective preprocessing and data augmentation to improve model generalization. Key Words: Retina Scan, Deep Learning, Convolutional To evaluate the model using standard performance metrics Neural Network, Eye Disease Prediction, Tele ophthalmology including accuracy, precision, recall, and F1-score.
1. INTRODUCTION
To propose future integration methods for clinical deployment and teleophthalmology systems.
Blindness and Visual impairment is identified as major health issues in public. According to World Health Organization (WHO), over 2.2 billion peoples are having vision impairment or blindness, with many of these cases treatable or preventable if they are early diagnosed. Retinal imaging using fundus photography is most effective diagnostic tools to detect eye diseases early.
4.LITERATURE REVIEW [1] This paper highlights the deep learning based automated system that uses OCT images to classify the retinal diseases. In order to identify the disease related to the problems four class classifications are used. These divisions rely on residual neural networks which are utilized in proposed classification algorithm. The study uses retinal OCT image dataset and 10fold cross validation procedure at patient level. The suggest method in the literature matched with classification accuracy of 0.973, sensitivity of 0.963 and specificity of 0.985 at B-scan level. The study also included qualitative assessment utilizing occlusion testing.
Moreover, manual inspections of retinal images are time consuming and depend heavily on expertise of ophthalmologists. In recent years, Deep learning (DL) technologies have revolutionized the medical imaging field in AI techniques. With high computational power and vast public datasets, DL models have achieved state-of-art performance in image classification tasks. The proposed project leverages this technology to build a robust and scalable solution for automated eye disease prediction using retina scans.
[2] Provided a thorough analysis of the ways in which deep learning and artificial intelligence (AI) are being used in ophthalmology. They talked about the efficacy of different deep learning architectures, like convolution neural networks (CNNs), in identifying eye conditions like glaucoma, age-related macular degeneration, and diabetic retinopathy. The authors underlined the potential of AI to improve diagnostic efficiency and accuracy in clinical settings and stressed the significance of sizable, annotated datasets for training strong AI models. The paper also
2. PROBLEM STATEMENT In spite of the availability of retinal imaging tools, the early detection of diseases remains limited with shortfall of accessibility to specialists especially in developing or rural
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 647