摘 要:针对医学领域中腺体细胞分割问题,提出基于改进 U-net 网络的腺体细胞图像分割算法,提供一种高可用性图像处理模型。该模型能够增强腺体细胞特征,减少信息丢失,借用 OpenCV 对腺体细胞轮廓进行颜色处理。采用 U-net 网络结合空洞残差模块,提高细胞分割精确度。文章方法在ISBI数据集评测,设计多组实验对比,验证可行性,实验结果Dice 系数达0.925,表明对腺体细胞图像分割算法存在较高应用价值。
关键词:腺体细胞分割;U-net 网络;图像处理;空洞残差模块;Dice 系数
DOI:10.19850/j.cnki.2096-4706.2022.013.023
中图分类号:TP391.4 文献标识码:A 文章编号:2096-4706(2022)13-0094-04
Application Research of Glandular Cell Image Segmentation Algorithm Based on Improved U-net Network
ZHOU Tao
(Guangzhou College of Commerce, Guangzhou 511363, China)
Abstract: Aiming at the problem of glandular cell segmentation in the medical field, a glandular cell image segmentation algorithm based on improved U-net network is proposed, which provides a high-availability image processing model. The model can enhance the glandular cell features, reduce the loss of information, and use OpenCV to color the glandular cell outline. The U-net network combined with the dilated residual module is used to improve the accuracy of cell segmentation. The method in this paper is evaluated on the ISBI data set, and comparison of multiple sets of experiments is designed to verify the feasibility. The experimental results show that the Dice coefficient reaches 0.925, which indicates that it has higher application value to the glandular cell image segmentation algorithm.
Keywords: glandular cell segmentation; U-net network; image processing; dilated residual module; Dice coefficient
参考文献:
[1] 林晓,邱晓嘉 . 图像分析技术在医学上的应用 [J]. 包头医学院学报,2005(3):311-314.
[2] 周贤善 . 医学图像处理技术综述 [J]. 福建电脑,2009,25(1):34+33.
[3] 国家癌症中心 . 中国肿瘤登记工作指导手册(2016) [M].北京:人民卫生出版社,2016.
[4] 陈爱斌,江霞 . 细胞分割算法研究方法综述 [J]. 电子世界,2011(15):76-79.
[5] 张晓宇,王彬,安卫超,等 . 基于融合损失函数的 3DU-Net++ 脑胶质瘤分割网络 [J]. 计算机科学,2021,48(9):187-193.
[6] 钟思华,郭兴明,郑伊能 . 改进 U-Net 网络的肺结节分割方法 [J]. 计算机工程与应用,2020,56(17):203-209.
[7] 蒋宏达,叶西宁 . 一种改进的 I-Unet 网络的皮肤病图像分割算法 [J]. 现代电子技术,2019,42(12):52-56.
[8] 何承恩,徐慧君,王忠,等 . 多模态磁共振脑肿瘤图像自动分割算法研究 [J]. 光学学报,2020,40(6):66-75.
[9] GURCAN MN,BOUCHERON LE,CAN A,e t al.Histopathological image analysis:a review [J].IEEE Rev Biomed Eng,2009,2:147-171.
[10] Chen Y P,Kalantidis Y,Li J S,et al. Multi-fiber networks for video recognition [C]//ECCV 2018.Cham:Springer, 2018:364-380.
作者简介:周涛(2000.07—),男,汉族,江西丰城人,网络工程师,本科,研究方向:机器学习、图像处理。