Optic disc detection and segmentation using saliency mask in retinal fundus images

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Optic disc (OD) detection is vital for diagnosing retinal conditions like diabetic retinopathy.
• Existing methods struggle with non-standard retinal image appearances.
• A robust OD segmentation technique is needed for improved clinical analysis.

Purpose of the Study:

• To develop a novel and robust algorithm for optic disc segmentation in color retinal fundus images.
• To accurately locate and delineate the optic disc, overcoming limitations of previous approaches.

Main Methods:

• A two-stage approach involving optic disc localization and segmentation.
• Localization includes preprocessing, vessel extraction/elimination, and geometric analysis.
• Segmentation combines multiple candidates to form a precise optic disc contour.

Main Results:

• Evaluated on 11 diverse databases, achieving high accuracy rates.
• Demonstrated 100% accuracy on multiple datasets (Chase, Drive, HRF, Drishti, Drions, etc.).
• Overall success rate of 99.80% and high specificity across databases.

Conclusions:

• The proposed method offers robustness and excellent performance, especially in critical retinal image cases.
• Achieves state-of-the-art results in optic disc detection and segmentation.
• Suitable for clinical use without requiring expert intervention per image.