基于Multiscale-Net的膝关节半月板分割方法  

Segmentation Method of Knee Meniscus Based on Multiscale-net

作　　者：王娟[1,2] 李传庚 张卿源 夏承遗 WANG Juan;LI Chuan-geng;ZHANG Qing-yuan;XIA Cheng-yi(School of Electrical Engineering and Automation,Tianjin University of Technology,Tianjin 300384,China;Tianjin Key Laboratory of Complex System Control Theory and Application,Tianjin 300384,China;School of Computer Science and Engineering,Tianjin University of Technology,Tianjin 300384,China)

机构地区：[1]天津理工大学电气工程与自动化学院,天津300384 [2]天津市复杂系统控制理论及应用重点实验室,天津300384 [3]天津理工大学计算机科学与工程学院,天津300384

出　　处：《计算机与现代化》2023年第5期111-116,共6页Computer and Modernization

基　　金：国家自然科学基金面上项目(61773286)。

摘　　要：膝关节半月板分割的精确度对于半月板撕裂等级的判别和诊断具有重大意义,为了提高分割精度,本文提出一种基于多尺度网络(Multiscale-Net)的膝关节半月板分割方法。该方法结合视觉几何组网16(Visual Geometry Group Network16,VGG16)的卷积层和池化层以及U-Net网络的解码器部分,将编码器和解码器相连的3×3卷积层替换为改进的空间卷积池化金字塔(Atrous Spatial Pyramid Pooling,ASPP)模块。最后在安徽医科大学第一附属医院提供的临床病人的真实数据集上进行验证并与U-Net、引入ASPP模块的U-Net等模型进行对比。实验结果表明本文方法的交并比(Intersection over Union,IoU)和DSC相似系数(Dice Similarity Coefficient,DSC)分别达到91.25%和94.89%。The accuracy of knee meniscus segmentation was of great significance to the discrimination and diagnosis of meniscus tear grade.In order to improve the segmentation accuracy,this paper proposed a knee meniscal segmentation method based on Multiscale-Net network.This method combined the convolution layer and pooling layer of visual geometry group network16 and the decoder part of U-Net,and it replaced the 3×3 convolution layer connected with the encoder and decoder with an improved atrous spatial pyramid pooling module.Finally,it was verified on the real data set of clinical patients provided by the first affili⁃ated hospital of Anhui medical university and compared with U-Net,U-Net with ASPP module introduced,and other models.The experimental results showed that the intersection over union and dice similarity coefficient of this method reached 91.25%and 94.89%respectively.

关 键 词：半月板图像分割 卷积神经网络 U-Net网络 空间卷积池化金字塔 VGG16 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]