Related Experiment Video
Updated: Aug 14, 2026
04:48
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
Published on: July 5, 2024
SDA-SwinNet: Swin-UNet with Dense Skip and Shift-ASPP for Retinal Vessel Segmentation
Jinghua Xiao1, Ming Zhao1,2,3, Rui Yang1
1School of Computer Science, Yangtze University, Jingzhou 434025, China.
Sensors (Basel, Switzerland)
|August 13, 2026
Summary
A new deep learning model, SDA-SwinNet, improves retinal artery and vein segmentation for ophthalmic diagnostics. This method enhances accuracy in segmenting complex, low-contrast retinal vessels.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Accurate retinal artery and vein segmentation is crucial for diagnosing various eye conditions.
- Existing methods struggle with the complex, variable, and low-contrast nature of retinal vasculature.
Purpose of the Study:
- To introduce SDA-SwinNet, a novel deep learning network designed to overcome challenges in retinal vessel segmentation.
- To evaluate the performance of SDA-SwinNet on standard datasets.
Main Methods:
- The study proposes SDA-SwinNet, utilizing Swin-UNet as a backbone.
- Key modifications include a Shift-ASPP module for multi-scale context, an HF-Bridge for feature fusion, and a fractal-constrained loss function.
Main Results:
- SDA-SwinNet achieved an F1-score of 73.13% on the DRIVE-AV dataset and 67.85% on the LES-AV dataset.
- The model demonstrated competitive segmentation accuracy for both arteries and veins.
- The network offers a favorable balance between segmentation performance and computational efficiency.
Conclusions:
- SDA-SwinNet effectively addresses the complexities of retinal vessel segmentation.
- The proposed model shows significant potential for improving automated ophthalmic diagnostic tools.
