Mean curvature and texture constrained composite weighted random walk algorithm for optic disc segmentation towards...

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Vision

Background:

• Accurate optic disc (OD) segmentation is crucial for glaucoma screening using fundus imaging.
• Challenges include subtle OD boundaries, blood vessel occlusion, and intensity inhomogeneity.

Purpose of the Study:

• To propose an improved random walk algorithm for robust optic disc segmentation.
• To enhance the accuracy of glaucoma screening through better OD segmentation.

Main Methods:

• An improved random walk algorithm incorporating mean curvature and Gabor texture energy features.
• A composite weight function was developed to compute edge weights.
• The method avoids issues like curve initialization and local energy minima inherent in deformable models.

Main Results:

• The proposed algorithm demonstrated robust and superior performance in optic disc segmentation.
• Validation across multiple datasets (DRIVE, DIARETDB1, DRISHTI-GS, MESSIDOR) confirmed effectiveness.
• Quantitative measures like Dice coefficient, sensitivity, and specificity were used to assess performance.

Conclusions:

• The improved random walk algorithm effectively addresses challenges in optic disc segmentation.
• This method offers a reliable approach for glaucoma screening via fundus imaging.
• The algorithm shows significant advantages over existing deformable model-based techniques.