Research Article
Deep Identification of Micro-Aneurysms for the Detection of Diabetic Retinopathy
@INPROCEEDINGS{10.4108/eai.19-12-2025.2362668, author={Praveen Modi and Harkaran Singh}, title={Deep Identification of Micro-Aneurysms for the Detection of Diabetic Retinopathy}, proceedings={Proceedings of the 1st International Conference on Computational Techniques in Data Science, IICTDS 2025, 19-20 December 2025, Chandigarh, India}, publisher={EAI}, proceedings_a={IICTDS}, year={2026}, month={11}, keywords={WCLAHE ANN SVM Over-Fitting Bi-Linear}, doi={10.4108/eai.19-12-2025.2362668} }- Praveen Modi
Harkaran Singh
Year: 2026
Deep Identification of Micro-Aneurysms for the Detection of Diabetic Retinopathy
IICTDS
EAI
DOI: 10.4108/eai.19-12-2025.2362668
Abstract
The detection of diabetic retinopathy (DR), using fundus images, helps the ophthalmologist to make decision about its severity level. These images may have various significance issues like dark brightness, blurring etc. This paper is focused on the effective identification of tiny red dots-micro aneurysms and improving the quality of fundus images. WCLAHE technique has been adopted to improve the contrast. Furthermore the bi-linear interpolation method has been applied to fine the irregular boundaries in the fundus images. These images have been effectively segmented to locate the region of interest (ROI) from which the statistical features have been computed. The issues of over fitting and bias have been effectively resolved in the improved support vector machine (I-SVM) model. The proposed solution has been tested on the collected dataset from IEEE Data Port, consist of 3,200 images, encompassing both DR and non DR cases. The performance metrics such as accuracy, F1-score, AUC have been used to analyze the results The proposed model has shown the better accuracy (96%), F1-Score (93.29%) sensitivity (92.13%), and specificity (94.49%) under 10-fold cross-validation as compared with the existing classification.