摘要
针对现有算法存在因视网膜血管尺寸微小和对比度低等造成细小血管分割缺失以及因病理区域造成血管过分割等问题,提出一种基于U型网络多尺度注意力细化视网膜分割算法。在编码和解码阶段使用改进的密集卷积模块充分提取血管的特征信息,提升特征的利用率。将不同尺度的编码层特征提取的结果拼接后,通过跳跃连接经双向注意力机制将特征增强后传递到解码层。在解码处引入空间细化模块进一步提取微小血管的空间信息,减少背景伪影,细化血管形态。该算法在公开数据集DRIVE和STARE上进行验证,其在评估指标准确率分别为0.9649和0.9663,灵敏度分别为0.8422和0.8050,特异性分别为0.9822和0.9880,AUC分别为0.9867和0.9895。
Aiming at the problems of unsegmented small blood vessels and over-segmented pathological areas due to the small size of retinal blood vessels and low contrast in existing algorithms,a multi-scale attention thinning retinal segmen-tation algorithm based on U-shaped network is proposed.First of all,the improved dense convolution module is used in the encoding and decoding stages to fully extract the feature information of the blood vessel,and improve the utilization of features.Secondly,the results of the four feature extractions of the coding layers of different scales are spliced,and then transferred to the decoding layer through skip connections.At the same time,a dual attention mechanism is intro-duced in the skipping connection and spatial refinement structure to spatially enhance the structure of the tiny blood ves-sels.Finally,the spatial refinement module is introduced in the decoding to further extract the spatial information of the tiny blood vessels and refine the distribution and shape of the blood vessels.The algorithm is verified on the public data sets DRIVE and STARE.The evaluation indicators ACC are 0.9649 and 0.9663,the sensitivity is 0.8422 and 0.8050,the specificity is 0.9822 and 0.9880,and the AUC is 0.9867 and 0.9895.
作者
梁礼明
陈鑫
余洁
周珑颂
LIANG Liming;CHEN Xin;YU Jie;ZHOU Longsong(School of Electrical Engineering and Automation,Jiangxi University of Science and Technology,Ganzhou,Jiangxi 341000,China)
出处
《计算机工程与应用》
CSCD
北大核心
2023年第6期212-220,共9页
Computer Engineering and Applications
基金
国家自然科学基金(51365017,61463018)
江西省自然科学基金面上项目(20192BAB205084)
江西省教育厅科学技术研究重点项目(GJJ170491)。
关键词
视网膜血管分割
空间细化
密集卷积
双向注意力
retinal vessel segmentation
spatial refinement
dense convolution
dual attention
