DAVS-NET: Dense Aggregation Vessel Segmentation Network for retinal vasculature detection in fundus images.

Area of Science:

• Ophthalmology
• Medical Imaging
• Computer Science

Background:

• Diabetic retinopathy (DR) is a leading cause of vision loss in diabetic patients.
• Early detection of DR relies on analyzing changes in retinal vasculature.
• Current deep learning methods for retinal vessel segmentation often require extensive parameters and struggle with image quality variations.

Purpose of the Study:

• To develop a novel deep learning model for accurate and efficient retinal blood vessel segmentation.
• To address the challenges of low-quality images and intensity variations in retinal datasets.
• To reduce the number of trainable parameters required for effective segmentation.

Main Methods:

• Introduction of the Dense Aggregation Vessel Segmentation Network (DAVS-Net), an encoder-decoder framework.
• Integration of edge information transfer from early encoder layers to the decoder for faster convergence.
• Evaluation on publicly available datasets: DRIVE, CHASE_DB1, and STARE.

Main Results:

• DAVS-Net achieved state-of-the-art segmentation accuracy on multiple retinal datasets.
• The proposed network demonstrates high performance with a significantly reduced number of trainable parameters.
• Effective handling of image quality variations and intensity differences was observed.

Conclusions:

• DAVS-Net offers a highly accurate and parameter-efficient solution for retinal blood vessel segmentation.
• This method can serve as a valuable tool for screening and diagnosing diabetic retinopathy.
• The approach shows promise for improving automated analysis of retinal images in clinical practice.