Ensemble selection for feature-based classification of diabetic maculopathy images.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Science

Background:

• Diabetic maculopathy (DM) is a leading cause of blindness globally.
• Early detection of DM is essential for timely intervention and vision preservation.
• Automated diagnostic tools can aid in the early identification of DM.

Purpose of the Study:

• To develop an automated decision system for classifying diabetic maculopathy (DM) fundus images.
• To categorize images into normal, clinically significant macular edema (CMSE), and non-clinically significant macular edema (non-CMSE).
• To extract, select, and utilize discriminatory textural features for accurate classification.

Main Methods:

• Extraction of both region-specific and global textural features from DM fundus images.
• Application of an ensemble classifier combining Hidden Naïve Bayes, Naïve Bayes, Sequential Minimal Optimization (SMO), and J48 classifiers.
• Utilizing a five-fold cross-validation approach for performance evaluation.

Main Results:

• Achieved an average classification accuracy of 96.7% in distinguishing DM severity levels.
• Demonstrated the effectiveness of textural feature extraction and ensemble classification for DM detection.
• Successfully classified DM fundus images into normal, CMSE, and non-CMSE categories.

Conclusions:

• The proposed automated system shows high accuracy in classifying diabetic maculopathy.
• Textural feature analysis combined with ensemble classification is a promising approach for early DM detection.
• This system can contribute to the early diagnosis and management of diabetic retinopathy to prevent vision loss.