摘要
针对超声心动图噪声大、纹理不均匀、边缘难分割的特点,提出一种基于DeepLabv3的多通道模型对左心室区域进行分割的方法。将DeepLabv3嵌入到堆叠式的模型中,分别学习前景特征(左心室特征)和背景特征,然后整合两个通道以产生最终的分割结果。实验结果表明,所提算法像素平均交并比(MIoU)比医学语义分割U-Net算法高出13.37%,在图像关键分割区域纹理不均匀的情况下,多通道DeepLabv3网络的分割效果明显好于仅使用DeepLabv3网络的分割效果。
Echocardiography has the characteristics of high noise,uneven texture and rough edges.This paper proposed a method for segmenting the left ventricular region based on the multi-channel model of DeepLabv3.DeepLabv3 was embedded into a stacked model,learning foreground features(left ventricular features)and background features,and then the two channels were integrated to produce the final segmentation results.The Mean Intersection over Union(MIoU)of this algorithm is 13.37%higher than that of the U-Net algorithm.
作者
薛飞
伍岳庆
姚宇
任伟
XUE Fei;WU Yueqing;YAO Yu;REN Wei(Chengdu Institute of Computer Application,Chinese Academy of Sciences,Chengdu Sichuan 610041,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《计算机应用》
CSCD
北大核心
2019年第S02期114-117,共4页
journal of Computer Applications
基金
四川省科技支撑计划重点研发项目(2017SZ0010)
四川省新一代人工智能重大专项(2018GZDZX0036)
