Design of Content Based Image Retrieval Scheme for Diabetic Retinopathy Images using Harmony Search Algorithm

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Science

Background:

• Diabetic Retinopathy (DR) is a complication of diabetes affecting retinal blood vessels.
• Early detection and screening are crucial for managing DR and preventing vision loss.
• Automated methods for retinal blood vessel segmentation are needed for efficient diagnosis.

Purpose of the Study:

• To develop an automated system for retinal blood vessel segmentation using an evolutionary algorithm.
• To design a Content Based Image Retrieval (CBIR) system for DR screening.
• To evaluate the performance of the developed CBIR system compared to conventional methods.

Main Methods:

• Retinal images were preprocessed to enhance vessel contrast.
• Blood vessels were segmented using Harmony Search Algorithm (HSA) combined with Otsu Multilevel Thresholding (MLT).
• A CBIR system was developed using segmented image features and Euclidean Distance for similarity matching.

Main Results:

• The HSA-MLT segmentation achieved high precision (96%) and recall (58%) in the CBIR system.
• The developed CBIR system demonstrated effective retrieval of similar retinal images.
• The automated system showed potential for computer-assisted diagnosis in DR screening.

Conclusions:

• The automated CBIR system using HSA-MLT segmentation is effective for diabetic retinopathy screening.
• This approach can assist physicians in clinical decision-making and research.
• Further development could enhance computer-assisted diagnosis for ophthalmological conditions.