DIABETIC RETINOPATHY DETECTION USING HAAR WAVELET TRANSFORM AND LENET CNN

Abstract

Diabetic Retinopathy (DR) is one of the most
serious complications of diabetes and a leading
cause of vision impairment among diabetic patients
worldwide. Early detection and continuous
monitoring are essential to prevent permanent
blindness and improve patient outcomes.
Traditional manual screening of retinal fundus
images by ophthalmologists is time-consuming,
costly, and prone to human error, especially when
dealing with large numbers of patients. To
overcome these limitations, this project proposes an
automated diabetic retinopathy detection system
using Haar Wavelet Transform (HWT) and LeNet
Convolutional Neural Network (CNN). The
proposed system enhances retinal image analysis
by combining image preprocessing, feature
extraction, and deep learning classification
techniques. Initially, retinal fundus images are
resized, normalized, and preprocessed to improve
image quality. Haar Wavelet Transform is applied
to extract significant retinal features such as blood
vessels, lesions, and texture variations. These
extracted features are then processed using the
LeNet CNN model for accurate classification of
retinal conditions into normal and hypertensive
retinopathy categories. In addition to retinal
analysis, the system also evaluates heart disease
risk using clinical parameters such as age, blood
pressure, cholesterol levels, and other
cardiovascular indicators through machine learning
algorithms. The system integrates both retinal
prediction and heart disease analysis to generate a
combined health risk assessment. A web-based
interface is developed to allow users to upload
retinal images and enter clinical data conveniently.
The proposed system improves diagnostic
accuracy, reduces manual workload, supports early
disease detection, and assists healthcare
professionals in making faster and more reliable
medical decisions.