Related Experiment Video
Updated: Jan 6, 2026
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
Enhanced retinal blood vessel segmentation via loss balancing in dense generative adversarial networks with quick
Daria Sandeep1, K Baranitharan2, A Padmavathi3
1Department of Information Technology, MLR Institute of Technology, Hyderabad, Telangana, India. sandeeplight97@mlrinstitutions.ac.in.
Insights
This study introduces an advanced framework for retinal blood vessel segmentation, achieving high accuracy in detecting fine vessels for diagnosing retinal diseases. The integrated approach enhances robustness for reliable clinical use.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Manual segmentation of retinal vessels is crucial for diagnosing conditions like diabetic retinopathy.
- Existing automated methods struggle with fine vessel segmentation and loss function optimization.
Purpose of the Study:
- To develop an integrated framework for enhanced retinal vessel segmentation accuracy and robustness.
- To improve automated detection of retinal vascular abnormalities for clinical applications.
Main Methods:
- Preprocessing with Quasi-Cross Bilateral Filtering (QCBF) for noise reduction.
- Feature extraction using Directed Acyclic Graph Neural Network with VGG16 (DAGNN-VGG16).
- Segmentation via Dense Generative Adversarial Network with Quick Attention Network (Dense GAN-QAN) and loss minimization using Swarm Bipolar Algorithm (SBA).
Main Results:
- Achieved high performance across three datasets (CHASE-DB1, STARE, DRIVE) with accuracy: 99.87%, F1-score: 99.82%.
- Demonstrated strong generalization and robustness with mean precision: 99.84%, recall: 99.78%, specificity: 99.87%.
Conclusions:
- The QCBF-DAGNN-VGG16-Dense GAN-QAN-SBA framework sets a new standard in retinal vessel segmentation.
- The approach effectively segments fine vessels and optimizes training, showing potential for clinical deployment in retinal disease diagnosis.
Purpose:
Manual segmentation of retinal blood vessels in fundus images has been widely used for detecting vascular occlusion, diabetic retinopathy, and other retinal conditions. However, existing automated methods face challenges in accurately segmenting fine vessels and optimizing loss functions effectively. This study aims to develop an integrated framework that enhances vessel segmentation accuracy and robustness for clinical applications.
Methods:
The proposed pipeline integrates multiple advanced techniques to address the limitations of current approaches. In preprocessing, Quasi-Cross Bilateral Filtering (QCBF) is applied to reduce noise and enhance vessel visibility. Feature extraction is performed using a Directed Acyclic Graph Neural Network with VGG16 (DAGNN-VGG16) for hierarchical and topologically-aware representation learning. Segmentation is achieved using a Dense Generative Adversarial Network with Quick Attention Network (Dense GAN-QAN), which balances loss and emphasizes critical vessel features. To further optimize training convergence, the Swarm Bipolar Algorithm (SBA) is employed for loss minimization.
Results:
The method was evaluated on three benchmark retinal vessel segmentation datasets-CHASE-DB1, STARE, and DRIVE-using sixfold cross-validation. The proposed approach achieved consistently high performance with mean results of accuracy: 99.87%, F1- score: 99.82%, precision: 99.84%, recall: 99.78%, and specificity: 99.87% across all datasets, demonstrating strong generalization and robustness.
Conclusion:
The integrated QCBF-DAGNN-VGG16-Dense GAN-QAN-SBA framework advances the state-of-the-art in retinal vessel segmentation by effectively handling fine vessel structures and ensuring optimized training. Its consistently high performance across multiple datasets highlights its potential for reliable clinical deployment in retinal disease detection and diagnosis.
