Localization and segmentation of optic disc in retinal images using circular Hough transform and grow-cut algorithm

Area of Science:

• Ophthalmology
• Medical Image Analysis
• Computer Vision

Background:

• Automated retinal image analysis is vital for early disease detection, including glaucoma and diabetic retinopathy.
• Accurate optic disc detection and segmentation are preliminary steps for computer-assisted glaucoma diagnostic systems.

Purpose of the Study:

• To present a robust methodology for optic disc detection and boundary segmentation in retinal images.
• To develop a foundational component for a computer-assisted glaucoma diagnostic system.

Main Methods:

• The proposed method utilizes morphological operations for image enhancement and pathology removal.
• Circular Hough transform is employed for approximating the optic disc center.
• The grow-cut algorithm is used for precise segmentation of the optic disc boundary.

Main Results:

• The method achieved a 100% optic disc detection success rate on most tested databases.
• High average spatial overlap percentages were recorded for optic disc boundary detection across multiple datasets (e.g., 87.93% on Messidor).
• The methodology demonstrated significant improvements over existing techniques for optic disc detection and boundary extraction.

Conclusions:

• The developed methodology offers a robust and accurate approach for optic disc detection and segmentation in retinal images.
• This technique serves as a significant advancement for computer-assisted diagnosis of glaucoma.
• The high success rates and improved performance indicate the clinical potential of this automated analysis tool.