Optic cup segmentation: type-II fuzzy thresholding approach and blood vessel extraction

Area of Science:

• Ophthalmology
• Medical Image Processing
• Computer Vision

Background:

• Optic cup segmentation is crucial for diagnosing optic nerve head diseases like glaucoma.
• The complex structure of the optic nerve head and its blood vessels presents significant challenges for automated segmentation.
• Existing methods often struggle with accuracy and require manual intervention.

Purpose of the Study:

• To develop and evaluate a novel, automated technique for segmenting the optic cup in two-dimensional (2D) fundus images.
• To leverage blood vessel characteristics for improved optic cup boundary detection.
• To assess the algorithm's accuracy and agreement with expert human segmentation.

Main Methods:

• Blood vessel extraction using top-hat transform and Otsu's segmentation to identify vessel kinks indicative of the cup boundary.
• Application of an interval type-II fuzzy entropy procedure for enhanced feature extraction.
• Hough transform for approximating the final optic cup boundary.
• Validation on a dataset of 550 fundus images with manual markings from six ophthalmologists.

Main Results:

• The algorithm achieved a cup detection accuracy of 78.2% (based on area and centroid) on 441 images.
• The algorithm demonstrated agreement with expert ophthalmologists in 356 out of 550 images.
• The highest inter-ophthalmologist agreement was observed among three experts in 398 out of 550 images.

Conclusions:

• The proposed technique offers a robust and automated approach for optic cup segmentation in fundus images.
• The method shows competitive performance compared to manual segmentation by ophthalmologists.
• This automated tool has the potential to aid in the early detection and monitoring of optic nerve head conditions.