A supervised blood vessel segmentation technique for digital Fundus images using Zernike Moment based features

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Accurate blood vessel segmentation is crucial for diagnosing various ocular pathologies.
• Existing methods face challenges in segmenting fine blood vessels and achieving high accuracy.

Purpose of the Study:

• To develop a supervised method for precise blood vessel segmentation using Zernike moment-based shape descriptors.
• To improve the detection of thin blood vessels for earlier pathological diagnosis.

Main Methods:

• A pixel-wise classification approach using an 11-D feature vector combining statistical and Zernike moment features.
• Optimal Zernike moment coefficients were selected for maximum differentiability between vessel and background pixels.
• An Artificial Neural Network (ANN) binary classifier was trained on manually selected points from the DRIVE dataset.

Main Results:

• Achieved high accuracies of 0.945 on the DRIVE dataset and 0.9486 on the STARE dataset.
• Outperformed existing supervised learning methods in blood vessel segmentation.
• Demonstrated superior performance in segmenting thinner blood vessels compared to previous techniques.

Conclusions:

• The proposed Zernike moment-based method offers a significant advancement in supervised blood vessel segmentation.
• Improved segmentation of thin vessels facilitates earlier detection of diseases.
• This method holds promise for enhanced diagnostic capabilities in ophthalmology and related fields.