Grading Of Diabetic Retinopathy Using Deep Learning
Diabetic retinopathy is a serious eye condition affecting individuals with diabetes and can lead to vision impairment or even blindness if not detected and treated in its early stages. In this project, we present a novel approach to automate the grading of diabetic retinopathy using deep learning techniques, implemented in Matlab. Our developed system leverages a Convolutional Neural Network (CNN) model architecture to achieve a remarkable accuracy rate of 92%, offering a reliable and efficient solution for diabetic retinopathy diagnosis. The proposed system consists of a multi-stage process. In the initial stage, retinal images are inputted into the system. The second stage encompasses preprocessing, which plays a pivotal role in enhancing the quality of the input images. It involves the removal of noise using a median filter and the improvement of image contrast through the Contrast Limited Adaptive Histogram Equalization (CLAHE) process. The heart of our system lies in the third module, where retinal images are subjected to classification based on the severity of diabetic retinopathy. Our CNN model, finely tuned with hyperparameters including epochs, learning rate, dropout rate, and optimizer (Stochastic Gradient Descent with Momentum – SGDM), is trained to make accurate predictions. It classifies retinal images into distinct grades, including ‘no apparent retinopathy’ (grade 0), ‘mild NPDR’ (grade 1), ‘moderate NPDR’ (grade 2), ‘severe NPDR’ (grade 3), and ‘Proliferative Diabetic Retinopathy’ (grade 4). This grading system aids in early diagnosis and timely intervention. To assess the performance of our proposed model, we employ a comprehensive set of evaluation metrics, including accuracy, error rate, precision, recall, specificity, F1-score, and Matthews Correlation Coefficient (MCC). These metrics provide a holistic evaluation of the system’s performance, ensuring its reliability and effectiveness in diabetic retinopathy grading. In conclusion, our project represents a significant advancement in the automated diagnosis of diabetic retinopathy, demonstrating the power of deep learning and CNNs in medical image analysis. With an impressive accuracy of 92%, our system holds great promise for improving the early detection and management of diabetic retinopathy, thereby enhancing the quality of life for individuals living with diabetes.
PROJECT OUTPUT VIDEO:
ALGORITHM / MODEL USED:
CNN Model Architecture.
- System : Pentium i3 Processor.
- Hard Disk : 500 GB.
- Monitor : 15’’ LED.
- Input Devices : Keyboard, Mouse.
- Ram : 8 GB.
- Operating system : Windows 10 Pro.
- Coding Language : MATLAB
- Tool : MATLABR2021A
Qingshan Hou, Peng Cao, Liyu Jia , Leqi Chen, Jinzhu Yang, and Osmar R. Zaiane, “Image Quality Assessment Guided Collaborative Learning of Image Enhancement and Classification for Diabetic Retinopathy Grading”, IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS, VOL. 27, NO. 3, MARCH 2023.