INFUSE 2025: International Conference on Frontiers of Unified Science and Exploration

Analysis and Detection of Multi-Lesion for Diabetic Retinopathy using CNN Based Approach

Not scheduled

20m

Conference Hall (School of Sciences, Bengaluru, India)

Oral Physical Sciences

Speaker

Dr Santosh Sivrajsingh Chowhan (College of Computer Science and Information Technology. Latur - 413531 (Maharashtra) India.)

Description

Abstract:
Diabetes is a global health concern affecting individuals across all age groups. Diabetic retinopathy (DR), a major ocular complication of diabetes, can lead to vision loss if not diagnosed and treated promptly. Traditional DR detection methods rely on manual examination by specialists, which is time-consuming and inconsistent. Key steps in DR diagnosis include retinal vasculature extraction and optic disc/fovea segmentation. Detecting lesions such as microaneurysms (MA), hemorrhages (HM), and exudates (EX) is essential for determining the DR stage.With advancements in deep learning, Convolutional Neural Network (CNN)-based methods have become prominent in DR research. This study presents a CNN-based framework for segmenting and classifying retinal lesions. A comprehensive literature review is conducted, and the proposed method is evaluated on publicly available datasets.
Keywords: Diabetic Retinopathy, CNN, Retinal Blood Vessel Segmentation, Lesion Detection