视网膜血管图像分割的尺度特征表示学习网络  

Scale feature representation learning network for retinal vessels image segmentation

作　　者：杨可欣 刘骊[1,2] 付晓东 刘利军[1,2] 彭玮 Yang Kexin;Liu Li;Fu Xiaodong;Liu Lijun;Peng Wei(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China;Computer Technology Application Key Laboratory of Yunnan Province,Kunming 650500,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500 [2]云南省计算机技术应用重点实验室,昆明650500

出　　处：《中国图象图形学报》2025年第3期855-869,共15页Journal of Image and Graphics

基　　金：国家自然科学基金项目(62262036,62362043);兴滇英才支持计划项目(KKXY202203008)。

摘　　要：目的针对视网膜血管图像分割中血管特征尺度多变、毛细血管细节丰富以及视杯视盘、病变等特殊区域干扰导致的表征不精确、分割误差大以及结果不准确等问题,提出一种视网膜血管图像分割的尺度特征表示学习网络,包括尺度特征表示、纹理特征增强和双重对比学习3个模块。方法首先,输入视网膜图像集中的图像,通过引入空间自注意力构建尺度特征表示模块,对视网膜血管进行分层尺度表征;然后,采用上下文信息引导的纹理滤波器对血管尺度特征进行纹理特征增强;最后,通过采样血管尺度特征和纹理增强特征,并定义联合损失进行双重对比学习,优化两种特征空间中视杯视盘、病变等特殊区域的血管。结果为了验证方法的有效性,在3个具有挑战性的数据集上进行对比实验,结果表明,构建的视网膜血管图像分割网络有助于准确表示血管尺度特征和纹理增强特征,能够较好地获得完整的视网膜毛细血管等特殊区域的血管分割结果。本文方法在DRIVE(digital retinal images for vessel extraction)数据集中较对比的大多数方法,Acc(accuracy)值平均提高了0.67%,Sp(specificity)值平均提高了0.48%;在STARE(structured analysis of the retina)数据集中较对比的大多数方法,Se(sensitivity)值平均提高了6.01%,Sp值平均提高了6.86%;在CHASE_DB1(child heart and health study in England)数据集中较对比的大多数方法,Se值平均提高了1.88%,F1(F1 score)值平均提高了1.98%。结论本文提出的视网膜血管图像分割网络,能精准分割多尺度血管、毛细血管和特殊区域的血管,有效辅助视网膜血管疾病诊断。Objective Retinal vessel image segmentation refers to the process of separating vessel pixels in a color fundus image from the background pixels.The morphology of retinal vessels is closely associated with various ophthalmic diseases and plays a crucial role in computer-aided diagnosis and smart medicine.Additionally,retinal vessel images provide important biological information that can be used as a basis for personal identification systems in the field of social security.Furthermore,segmented retinal vessel images can serve as a priori for other anatomical sites,such as the macula.Currently,retinal image segmentation methods can be categorized into traditional and deep learning methods.Existing methods for retinal vessel image segmentation demonstrate good performance in segmenting large-scale vessels,primarily due to the ease of capturing features related to these prominent structures.Particularly,U-Net can effectively handle the complicated anatomical semantics involved in retinal vessel segmentation tasks,fusing adjacent-level features to learn additional local and global semantic information for highly accurate segmentation.Although remarkable progress has been made in retinal vessel segmentation with the advancement of deep learning,several challenging issues remain.First,current methods do not adequately represent vessels feature at multiple scales,resulting in poor segmentation results for retinal vessels with large differences in size and shape.Second,thin vessels,particularly those located at the ends of extremely low-contrast branches,are easily missed by current methods,resulting in incomplete vessel segmentation.Additionally,the medical semantics surrounding retinal vessels are complex.Specific regions,such as the optic cup,optic disc,and lesions,can interfere with vessel segmentation and seriously affect the accuracy of retinal vessel segmentation.Moreover,most images in the STARE dataset have severe lesions,and the information in different datasets notably varies,resulting in lower sensitivity of

关 键 词：视网膜血管图像分割 尺度特征表示 纹理特征增强 纹理滤波器 双重对比学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]