Convolutional neural network-based sea lion optimization algorithm for the detection and classification of diabetic...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Diabetic retinopathy (DR) is a leading cause of preventable blindness, damaging retinal blood vessels.
• Early detection of DR is vital to prevent irreversible vision loss, but manual diagnosis is time-consuming and prone to errors.
• Existing automated DR detection models often suffer from accuracy issues, high dimensionality, and computational complexity.

Purpose of the Study:

• To develop an accurate and efficient automated system for detecting diabetic retinopathy.
• To address the limitations of current DR detection models, including accuracy, data handling, and computational cost.
• To propose a novel deep learning approach for classifying DR severity from retinal fundus images.

Main Methods:

• Retinal images undergo preprocessing, including cropping for noise reduction and segmentation using a modified level set algorithm.
• An Aquila optimizer is utilized for feature extraction from segmented images.
• A convolutional neural network-oriented sea lion optimization (CNN-SLO) algorithm is proposed for DR classification.

Main Results:

• The CNN-SLO algorithm effectively classifies retinal images into five categories: healthy, mild, moderate, severe, and proliferative DR.
• The proposed method aims to overcome the shortcomings of existing DR detection models.
• Experimental validation on Kaggle datasets demonstrates the system's performance.

Conclusions:

• The developed CNN-SLO algorithm offers a promising solution for accurate and efficient diabetic retinopathy detection.
• The study highlights the potential of optimized deep learning models in medical image analysis.
• Further investigation using diverse datasets and evaluation metrics is recommended to validate the system's robustness.