Department of Electronic and Information Engineering, Shantou University, 515063, Guangdong, China.
Key Lab of Digital Signal and Image Processing of Guangdong Province, Shantou University, 515063, Guangdong, China.
Med Biol Eng Comput. 2023 Jul;61(7):1745-1755. doi: 10.1007/s11517-023-02806-1. Epub 2023 Mar 10.
Automated and accurate segmentation of retinal vessels in fundus images is an important step for screening and diagnosing various ophthalmologic diseases. However, many factors, including the variations of vessels in color, shape and size, make this task become an intricate challenge. One kind of the most popular methods for vessel segmentation is U-Net based methods. However, in the U-Net based methods, the size of the convolution kernels is generally fixed. As a result, the receptive field for an individual convolution operation is single, which is not conducive to the segmentation of retinal vessels with various thicknesses. To overcome this problem, in this paper, we employed self-calibrated convolutions to replace the traditional convolutions for the U-Net, which can make the U-Net learn discriminative representations from different receptive fields. Besides, we proposed an improved spatial attention module, instead of using traditional convolutions, to connect the encoding part and decoding part of the U-Net, which can improve the ability of the U-Net to detect thin vessels. The proposed method has been tested on Digital Retinal Images for Vessel Extraction (DRIVE) database and Child Heart and Health Study in England Database (CHASE DB1). The metrics used to evaluate the performance of the proposed method are accuracy (ACC), sensitivity (SE), specificity (SP), F1-score (F1) and the area under the receiver operating characteristic curve (AUC). The ACC, SE, SP, F1 and AUC obtained by the proposed method are 0.9680, 0.8036, 0.9840, 0.8138 and 0.9840 respectively on DRIVE database, and 0.9756, 0.8118, 0.9867, 0.8068 and 0.9888 respectively on CHASE DB1, which are better than those obtained by the traditional U-Net (the ACC, SE, SP, F1 and AUC obtained by U-Net are 0.9646, 0.7895, 0.9814, 0.7963 and 0.9791 respectively on DRIVE database, and 0.9733, 0.7817, 0.9862, 0.7870 and 0.9810 respectively on CHASE DB1). The experimental results indicate that the proposed modifications in the U-Net are effective for vessel segmentation. The structure of the proposed network.
自动且准确地分割眼底图像中的视网膜血管是筛选和诊断各种眼科疾病的重要步骤。然而，许多因素，包括血管在颜色、形状和大小上的变化，使得这项任务变得非常复杂。一种最受欢迎的血管分割方法是基于 U-Net 的方法。然而，在基于 U-Net 的方法中，卷积核的大小通常是固定的。因此，单个卷积操作的感受野是单一的，不利于分割各种厚度的视网膜血管。为了解决这个问题，本文采用自校准卷积代替传统卷积来替换 U-Net，这可以使 U-Net 从不同的感受野中学习到有区别的表示。此外，我们提出了一种改进的空间注意力模块，而不是使用传统的卷积，来连接 U-Net 的编码部分和解码部分，从而提高 U-Net 检测细血管的能力。所提出的方法已经在 Digital Retinal Images for Vessel Extraction (DRIVE) 数据库和 Child Heart and Health Study in England Database (CHASE DB1) 上进行了测试。用于评估所提出方法性能的指标是准确性 (ACC)、灵敏度 (SE)、特异性 (SP)、F1 分数 (F1) 和接收器工作特征曲线下的面积 (AUC)。所提出的方法在 DRIVE 数据库上的 ACC、SE、SP、F1 和 AUC 分别为 0.9680、0.8036、0.9840、0.8138 和 0.9840，在 CHASE DB1 上的分别为 0.9756、0.8118、0.9867、0.8068 和 0.9888，优于传统 U-Net（U-Net 在 DRIVE 数据库上的 ACC、SE、SP、F1 和 AUC 分别为 0.9646、0.7895、0.9814、0.7963 和 0.9791，在 CHASE DB1 上的分别为 0.9733、0.7817、0.9862、0.7870 和 0.9810）。实验结果表明，所提出的 U-Net 改进对于血管分割是有效的。所提出网络的结构。
