DenseNet 121 Framework For Automatic Feature Extraction Of Diabetic Retinopathy Images

Abstract

Automatic feature extraction from medical images is one of the most important aspects of early disease detection and treatment. This work presents a novel method for autonomously extracting features from diabetic retinopathy images using the DenseNet-121 framework. Diabetes frequently has a side effect called diabetic retinopathy, which requires early detection to protect vision and enhance patient outcomes. Typical methodologies for feature extraction from medical photographs usually struggle to capture the intricate details and patterns shown in diabetic retinopathy imagery. To solve these challenges and facilitate a more comprehensive analysis of retinopathy images, we propose to use the widely recognized DenseNet-121 architecture, which is well known for its deep feature extraction and feature reuse capabilities. The test results on a large sample of diabetic retinopathy images validate the