【摘要】 新冠病毒2019(COVID-19)在过去一年内严重威胁人类生命和健康。全球经济、教育、交通等方面都受到了影响。为了尽快解决新冠病毒引起的问题，快速准确判断人们是否感染是非常重要的。以U-Net为基本框架，采用多尺度特征提取模块，并在反卷积过程中加入位置信息，将图像的全局信息和局部信息充分结合，提出改进模型并应用于COVID-19 CT图像集。上述模型通过深度学习自动分割COVID-19 CT图像的左右肺、病灶和背景四部分。最后对分割结果进行了评价，达到了预期的效果。有助于医务人员快速识别感染部位。更多还原

【Abstract】 Corona Virus Disease 2019(COVID-19) has seriously threatened human life and health in the past year. The global economy, education, transportation, and other aspects have been affected. In order to solve the problems caused by COVID-19 as soon as possible, it is very important to quickly and accurately determine whether people are infected. In this paper, we take U-Net as the basic model, adopt a multi-scale feature extraction module and add position information in the deconvolution process to fully combine the global information and local information of the image. We propose a model suitable for the COVID-19 CT image sets. The model can automatically segment four parts(left and right lung, disease, and background) of COVID-19 CT images through deep learning. The segmentation results are evaluated and the expected results are achieved. It is helpful for medical staff to quickly identify the infection area.更多还原