Related Experiment Video
Updated: May 19, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
AI-driven saliency-guided retinal vessel segmentation framework for sustainable digital pathology
Rajib Guha Thakurta1, Mohammed E Seno2, Masood Ur Rehman3
1School of Computer Science and Applications, REVA University, Bangalore, India.
Frontiers in Medicine
|May 18, 2026
Summary
This study introduces SGB-Net, an AI framework for accurate retinal blood vessel segmentation. It improves detection of thin vessels and boundary continuity, aiding early disease diagnosis.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Ophthalmology
Background:
- Accurate retinal blood vessel segmentation is crucial for diagnosing ophthalmic and systemic diseases.
- Challenges include low contrast, complex vessel geometry, and artifacts, especially affecting thin vessels.
Purpose of the Study:
- To develop an AI-driven saliency-guided boundary refinement framework (SGB-Net) for enhanced retinal vessel segmentation.
- To improve the accuracy and robustness of automated retinal image analysis.
Main Methods:
- Proposed SGB-Net integrates a progressive boundary refinement (BR) module.
- Employs a feature-guided encoder-decoder network with scale-adaptive (SA) and attention enhancement (AE) modules.
- Evaluated on DRIVE, STARE, and CHASE_DB1 datasets.
Main Results:
- Achieved superior segmentation performance with Dice scores up to 98.30% and AUC up to 0.9899.
- Demonstrated improved preservation of thin vessels and boundary continuity.
- Reduced false positives in complex imaging conditions.
Conclusions:
- SGB-Net effectively addresses limitations in retinal vessel segmentation through boundary refinement and advanced feature learning.
- The framework shows robustness to noise and pathological variations, suitable for digital pathology.
- Potential for broader applications in automated retinal analysis and other medical imaging modalities.