Enhanced Retinal Vessel Segmentation Using U-Net Framework

The key step in retinal imaging is blood vessel augmentation, which is important for early diagnosing eye conditions such as glaucoma, macular edema, diabetic retinopathy, and more. Deep learning has been used in the segmentation process for the last few years due to its high accuracy and efficiency. We have proposed a convolutional neural network based on U-net to segment blood vessel fundus images in this paper. The publicly available DRIVE and CHASE DB1 datasets were used to train the model and for testing, the HRF dataset was used. For the DRIVE dataset, the research findings show an average accuracy of 93.2 percent.