Abstract:

The study proposes an approach for classifying diabetic retinopathy using hybrid neural network and an entropy-based feature extraction algorithm. The traditional method of diagnosis is sometimes prone to misinterpretation since pictures representing various disease stages are usually identical. In order to solve this issue, the proposed method uses a discrete wavelet transform-based entropy-based feature extraction strategy to improve the appearance of medical pictures and make subtle features more noticeable. To efficiently categorize images of diabetic retinopathy, a hybrid neural network was developed. Two datasets are used to validate the proposed approach. The effectiveness of the approach is validated by many experiments. The model also includes a classifier that takes an input image, generates a predicted label and an actual label, and also recommends any necessary safety precautions.

Description:

Diabetic Retinopathy, Hybrid Neural Network, Entropy based Feature extraction