Modified pixel level snake using bottom hat transformation for evolution of retinal vasculature map

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Retinal blood vessel changes are indicators of various pathological disorders that can lead to blindness.
• Accurate extraction of the retinal vasculature map is crucial for analyzing these pathologies.
• Current methods face challenges in precise vessel segmentation from retinal fundus images.

Purpose of the Study:

• To present an unsupervised method for accurate retinal vasculature map extraction from fundus images.
• To develop a technique for analyzing retinal vascular attributes to aid in diagnosing eye conditions.
• To improve the detection of blood vessels in both normal and pathological retinal images.

Main Methods:

• Utilized Black Top-Hat (BTH) transformation combined with a Modified Pixel Level Snake (MPLS) algorithm.
• Employed bimodal masking for retinal image mask extraction, followed by adaptive segmentation and global thresholding.
• Applied MPLS with external, internal, and balloon potentials for contour evolution in four cardinal directions.

Main Results:

• Achieved high performance metrics on the DRIVE database: 76.96% sensitivity, 98.34% specificity, and 96.30% accuracy.
• Demonstrated effectiveness on pathological images with an average sensitivity of 70.80%, specificity of 96.40%, and accuracy of 94.41%.
• The method provides a simple and accurate approach for vasculature detection.

Conclusions:

• The proposed unsupervised method accurately extracts retinal vasculature maps from normal and pathological fundus images.
• This technique can assist in assessing various retinal vascular attributes, contributing to early disease detection.
• The MPLS algorithm based on BTH transformation offers a promising tool for ophthalmological image analysis.