International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 06 | Jun 2025
p-ISSN: 2395-0072
www.irjet.net
Explainable Diabetic Retinopathy Detection using Deep Learning Pranav Vinod Chaudhari1, Devyani Prakash Badgujar2, Mandar Ravindra Visave3, Akanksha Bharat Chaudhari4, Manisha Shantaram Patil5 1Student, Dept. of AIML, R. C. Patel Institute of Technology, Shirpur, Maharashtra, India 2Student, Dept. of AIML, R. C. Patel Institute of Technology, Shirpur, Maharashtra, India 3Student, Dept. of AIML, R. C. Patel Institute of Technology, Shirpur, Maharashtra, India
4Student, Dept. of AIML, R. C. Patel Institute of Technology, Shirpur, Maharashtra, India 5Assistant Professor, Dept. of AIML, R. C. Patel Institute of Technology, Shirpur, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Diabetic Retinopathy (DR) is a critical
As the number of patients with diabetes increases, the number of retinal images produced by screening programs will also increase, placing a large burden on medical experts and healthcare systems. This could be alleviated with an automated system. Studies show that automated systems based on deep learning neural networks can achieve high sensitivity and specificity in detecting referable diabetic retinopathy. Other referable eye complications such as diabetic macular edema and glaucoma have also been explored [3].
complication of diabetes that leads to progressive vision impairment and, if left untreated, permanent blindness. With the increasing prevalence of diabetes worldwide, early and accurate detection of DR has become essential for timely intervention. Traditional manual screening methods are timeconsuming, prone to human error and require skilled ophthalmologists, which limits their scalability. Recent advances in deep learning have enabled the develop ment of automated systems that can detect and classify DR from retinal fundus images with high accuracy. However, most of these models function as ”black boxes” and lack transparency, making their predictions difficult to interpret in clinical settings. This paper presents a comprehensive and explainable DR detection pipeline using Convolutional Neural Network (CNN) architectures such as VGG16, Xception, ResNet50 and Sequential CNN. A publicly available real-world dataset of labeled fundus images is used for training and evaluation. Through comparative analysis, the performance of each model is assessed based on key metrics including accuracy, precision, recall and F1-score. The ultimate goal is to aid ophthalmologists with a reliable, accurate and interpretable diagnostic tool, while also laying the groundwork for the future incorporation of explainable AI techniques such as Grad-CAM and synthetic image generation using GANs.
Artificial Intelligence (AI), especially deep learning, has become a powerful ally in healthcare. Convolutional Neural Networks (CNNs) can automatically classify retinal images and detect DR stages. Despite their potential, these models often lack transparency in their predictions, which can be a barrier to clinical trust. This research proposes a deep learning pipeline for DR detection using various CNN architectures. We emphasize ac curacy, robustness and the foundation for future interpretability enhancements.
2. DATASET AND PREPROCESSING We used a fundus image dataset containing 3630 labeled images across five DR classes: No DR, Mild, Moderate, Severe and Proliferative DR.
Key Words - Diabetic Retinopathy, Deep Learning, CNN, Medical Imaging, Transfer Learning
2.1 Preprocessing
1.INTRODUCTION By 2040, approximately 600 million people are predicted to have diabetes and one third are expected to have diabetic retinopathy [1]. Diabetes reduces life expectancy by five to 10 years [2]. Diabetic retinopathy is the most common microvascular complication in diabetes [3]. It affects people with diabetes and is the result of prolonged high blood glucose levels that damage retinal blood vessels. Early detection can prevent vision loss, but access to ophthalmologists is often limited. Retinal imaging is the most widely used method due to its high sensitivity in detecting retinopathy [3].
© 2025, IRJET
|
Impact Factor value: 8.315
•
All images were resized to 224 × 224 pixels.
•
Pixel intensities were normalized to [0, 1].
•
Data augmentation (rotation, flip, brightness) was used to balance classes and reduce overfitting.
•
One-hot encoding was used for the labels.
2.1 Dataset Description We used the publicly available APTOS 2019 Blindness Detection dataset collected by Aravind Eye Hospital, India [4]. It originally had 3630 Images, but after dataset balancing it now contains 4530 labeled images rated for DR severity on a scale of 0 to 4. Distribution: 1016 (No DR), 840 (Mild), 999
|
ISO 9001:2008 Certified Journal
|
Page 440