ISSN: 2578-465X
Authors: Carolyn YTW*, Timing L, Tin LW and Henry HWL
Retinal vessel segmentation plays a crucial role in the automated examination of fundus images for screening and diagnosing diabetic retinopathy, a common complication of diabetes leading to sudden vision loss. Automated segmentation of retinal vessels can detect these changes more efficiently and accurately compared to manual assessment by an ophthalmologist. The proposed method aims to precisely identify blood vessels in retinal images while simplifying the segmentation process and reducing computational complexity. This approach can enhance the accuracy and reliability of retinal image analysis, aiding in the diagnosis of various eye diseases. The Attention Gated U-Net architecture is a key component in retinal image segmentation for retinal pathologies like diabetic retinopathy, showing promising results in improving segmentation accuracy, especially in scenarios with limited training data and ground truth. This method involves incorporating an attention mechanism into the U-Net to focus on relevant regions of the input image, enhancing the performance of semantic segmentation models. Extensive experiments conducted on a retinal segmentation dataset demonstrated that the proposed approach outperformed existing methods in terms of performance.
Keywords: Artificial Intelligence; Image Segmentation; Retinal Blood Vessel; Colour Fundus Photograph