GLCA Detection of Macular Age Affiliated Degeneration Using Ensemble Tree Classification in OCT Images Scan

¹ Department of Electronics and Communication Engineering, Annai Mira College of Engineering and Technology, Arapakkam, Ranipet -632517, Tamil Nadu, India.
² Department of Computer Science and Engineering, Annai Mira College of Engineering and Technology, Arapakkam, Ranipet- 632517, Tamil Nadu, India.

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Age-related macular degeneration (AMD) detection in OCT images is done utilizing GLCA texture features, optimized through genetic algorithms, and classified with an ensemble tree model for precise detection. Constraints include the use of texture alone, parameter sensitivity, requirement of large datasets, vulnerability to OCT noise, and lower clinical interpretability. This research introduces new algorithmic methods for improving the quality of Optical Coherence Tomography (OCT) images for the identification of age-related macular degeneration (AMD). Integrated into an automated AMD detection system, the proposed techniques enhance the identification of abnormalities in OCT scans, enabling correct classification of healthy and diseased macular tissues. The method combines texture analysis, statistical evaluation, and genetic algorithms to find the best features. A genetic algorithm is employed to feature selection through finding the most representative attributes, using in-depth texture descriptors of the Gray Level Co-occurrence Array (GLCA) as well as statistical parameters. GLCA features are derived at four orientations (0°, 45°, 90°, and 135°), and an ensemble tree model created based on optimized feature set. The model shows extreme efficacy, having an error rate of 0% with chosen features, 2.0% with all features, and 7.0%without features. Entire system has 100% model development accuracy on all cases, and evaluation accuracy of 91.358%, 96.103%, and 100% for no features, all features, and selected features, respectively.