Retinal Vascular Image Segmentation Using Improved UNet Based on Residual Module

Abstract

In recent years, deep learning technology for clinical diagnosis has progressed considerably, and the value of medical imaging continues to increase. In the past, clinicians evaluated medical images according to their individual expertise. In contrast, the application of artificial intelligence technology for automatic analysis and diagnostic assistance to support clinicians in evaluating medical information more efficiently has become an important trend. In this study, we propose a machine learning architecture designed to segment images of retinal blood vessels based on an improved U-Net neural network model. The proposed model incorporates a residual module to extract features more effectively, and includes a full-scale skip connection to combine low level details with high-level features at different scales. The results of an experimental evaluation show that the model was able to segment images of retinal vessels accurately. The proposed method also outperformed several existing models on the benchmark datasets DRIVE and ROSE, including U-Net, ResUNet, U-Net3+, ResUNet++, and CaraNet.

DeepVasNet: Retinal Vasculature and FAZ Segmentation in OCTA with EfficientNet-Enhanced U-Net

“…Unlike several existing techniques [1,5,11,[14][15][16][17][18][19], the proposed work separately delineates the different blood vessels (namely, artery, vein and capillary) and FAZ. This can be used to detect anomalies like artery and vein occlusion.…”

“…The skip connections that connect the corresponding parts of the encoder and decoder help incorporate multi-level feature information. Many previous OCTA image segmentation approaches [5,11,13,14,[16][17][18] have preferred the U-Net as their baseline model considering its ease of implementation and performance efficiency.…”

“…Various methods are available in the state-of-the-art literature for retinal blood vessels and FAZ segmentation. This includes delineation of the retinal vasculature alone [1,4,11,16] or combined retinal vessel and FAZ detection [5,14,15,[17][18][19]. Some studies further segment the retinal blood vessels into arteries, veins, and capillaries [2,8,20].…”

“…Several subsequent works utilized the ROSE dataset for OCTA segmentation and classification. One such example is the research of Huang et al [16]. Their algorithm involved a modified U-Net with residual blocks for better feature extraction and additional skip connections for integrating the features at multiple levels.…”

Multi-Layer Preprocessing and U-Net with Residual Attention Block for Retinal Blood Vessel Segmentation

“…Recently, DL presented many models for segmenting retinal fundus images, such as convolutional neural networks (CNN), fully convolutional networks (FCN), and encoder-decoder-based models, i.e., U-Net [14][15][16]. The U-Net and its variant architectures, such as U-Net++ and residual U-Net, prove their efficiency when compared with other DL models because of their accuracy and a small number of parameters during the training process [17,18]. Preprocessing of retinal images is a highly significant task before segmentation for increasing the accuracy of the segmentation and training process.…”