Related Experiment Video
Updated: May 5, 2026
13:19
Deep Neural Networks for Image-Based Dietary Assessment
Published on: March 13, 2021
11.0K
Hybrid deep learning framework for diabetic retinopathy classification with optimized attention AlexNet.
1Department of Computer Science and Engineering, Mar Ephraem College of Engineering and Technology, Elavuvilai, Tamilnadu, India.
Computers in Biology and Medicine
|March 28, 2025
Summary
This study introduces an advanced deep learning model for automated diabetic retinopathy (DR) classification. The model achieves high accuracy in detecting DR severity from retinal fundus images, improving early diagnosis.
Area of Science:
- Ophthalmology and Medical Imaging
- Artificial Intelligence in Healthcare
- Computational Pathology
Background:
- Diabetic retinopathy (DR) is a leading cause of vision loss, necessitating accurate and efficient detection methods.
- Manual DR assessment is time-consuming and prone to human error, highlighting the need for automated solutions.
- Deep learning (DL) shows promise for enhancing the accuracy and efficiency of DR classification.
Purpose of the Study:
- To develop and evaluate an automated system for classifying diabetic retinopathy severity using advanced deep learning.
- To improve the accuracy and efficiency of DR detection compared to manual methods.
- To leverage metaheuristic optimization for enhanced classification performance in retinal fundus images.
Main Methods:
- A four-stage automated DR classification pipeline was proposed, including pre-processing (green channel conversion, CLAHE, Gaussian filtering).
- Fundus lesions were segmented using Fuzzy Possibilistic C Ordered Means (FPCOM).
- Classification was performed by an Attention AlexNet model optimized with the Improved Nutcracker Optimizer (At-AlexNet-ImNO).
Main Results:
- The At-AlexNet-ImNO model achieved high performance on benchmark datasets (APTOS-2019 and EyePacs).
- Accuracy, precision, and recall rates exceeded 98% on both datasets, demonstrating robust classification capabilities.
- The ImNO optimizer effectively enhanced the performance of the At-AlexNet model by optimizing weights and hyperparameters.
Conclusions:
- The proposed automated DR classification system, utilizing metaheuristic-optimized deep learning, demonstrates significant potential for accurate and efficient DR grading.
- This approach offers a promising tool for early detection and management of diabetic retinopathy, potentially reducing irreversible vision loss.
- The study validates the effectiveness of combining advanced DL architectures with metaheuristic optimization for complex medical image analysis tasks.
