基于优化U-Net神经网络模型在医学图像分割的应用  

Application of Medical Image Segmentation Based on Optimized U-Net Neural Network Model

作　　者：张筱旭 邵英龙 严孟慧 王健庆[1] ZHANG Xiaoxu;SHAO Yinglong;YAN Menghui;WANG Jianqing(Zhejiang Chinese Medical University,Hangzhou 310053,China)

机构地区：[1]浙江中医药大学,浙江杭州310053

出　　处：《现代信息科技》2025年第4期47-52,共6页Modern Information Technology

摘　　要：医学图像是临床诊断的重要参考,如何快速且准确地分割出医学图像中的病灶区域,受到了人们的广泛关注。当前,利用深度学习进行图像处理已成为主流,医学图像分割因其独特的应用场景,成为深度学习在图像处理领域应用的成功范例。U-Net网络凭借其特有的U型结构,在医学图像分割领域取得了不错的性能,但该网络仍存在精度不够高等问题。文章对基于优化U-Net模型的医学图像自动分割方法展开研究,将CBAM(Convolutional Block Attention Module)和SE(Squeeze-and-Excitation)模块与U-Net网络结构相结合,实现了对人体器官的高度准确分割。在眼球数据集上的实验结果表明,优化后的U-Net网络相较于单纯的U-Net网络,准确率更高(0.905)。该研究具有重要的临床应用前景,能够对人体器官、病变区域等目标进行有效分割,为医疗实践带来积极影响。Medical images are important references for clinical diagnosis.How to segment the lesion areas in medical images quickly and accurately has received extensive attention.At present,the use of Deep Learning for image processing has become the mainstream.Medical image segmentation has become a successful example of Deep Learning in the field of image processing due to its unique application scenarios.With its unique U-shaped structure,the U-Net network has achieved good performance in the field of medical image segmentation,but the network still has problems such as insufficient accuracy.This paper studies the automatic segmentation method of medical images based on the optimized U-Net model.The CBAM and SE modules are combined with the U-Net network structure to achieve highly accurate segmentation of human organs.The experimental results on the eyeball dataset show that the optimized U-Net network has higher accuracy (0.905) than the simple U-Net network.This study has important clinical application prospects,which can effectively segment human organs,lesion areas and other targets,and has a positive impact on medical practice.

关 键 词：U-Net神经网络 图像分割 医学图像 注意力机制 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]