Automated segmentation of optic disc using statistical region merging and morphological operations

Area of Science:

• Ophthalmology
• Medical Image Analysis
• Computer Vision

Background:

• Accurate optic disc (OD) segmentation is essential for diagnosing early-stage retinal diseases.
• Automated segmentation is challenging due to ambiguous OD boundaries in many retinal images.

Purpose of the Study:

• To propose and evaluate a novel method for automated optic disc segmentation.
• To address the challenges posed by ambiguous optic disc boundaries in retinal imaging.

Main Methods:

• A new method for optic disc segmentation utilizing statistical region merging and morphological operations.
• Validation on diverse standard datasets: MESSIDOR, DIARETDB1, DIARETDB0, and DRIONS-DB.

Main Results:

• High average overlap ratios achieved: 91.35% (DIARETDB1), 88.80% (DRIONS-DB), 86.60% (DIARETDB0), and 89.68% (MESSIDOR).
• Exceptional average accuracies recorded: 99.68% (DIARETDB1), 99.89% (DRIONS-DB), 99.69% (DIARETDB0), and 99.93% (MESSIDOR).

Conclusions:

• The proposed method demonstrates superior performance in optic disc segmentation compared to existing algorithms.
• The technique offers a robust and accurate solution for automated OD segmentation in clinical applications.