融合SE和多尺度卷积的轻量级CT图像分类模型  

A lightweight CT image classification model for fusion SE and multi-scale convolution

作　　者：魏榕剑 邵剑飞[1] 温剑 冯宇航 叶榕 WEI Rong-jian;SHAO Jian-fei;WEN Jian;FENG Yu-hang;YE Rong(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,Yunnan,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,云南昆明650500

出　　处：《云南大学学报（自然科学版）》2023年第3期583-589,共7页Journal of Yunnan University(Natural Sciences Edition)

基　　金：国家自然科学基金(61732005).

摘　　要：针对医生主观因素会影响COVID-19(新型冠状病毒感染)和CAP(社区获得性肺炎)类型计算机断层(Computed Tomography,CT)图像诊断准确性的问题,提出一种基于MobileNetV2网络改进的MobileNetV2-SELN结构.首先,对MobileNetV2结构中的Block模块进行改进,添加SE块和尺度注意力机制,并引入全连接层和全局池化以便于获取多尺度特征;然后,针对COVID-19图像和CAP图像相似度大的特点,用GroupNorm替代BatchNorm2d,使模型能更好地获取肺炎特征;最后,使用SGD优化器对模型进行优化.实验结果表明,提出的模型的分类准确率更高.To address the problem that physician subjective factors can affect the Computed Tomography images diagnostic accuracy of COVID-19(novel coronavirus infection)and CAP(Community-Acquired Pneumonia)types,a MobileNetV2-SELN structure is proposed based on the improved MobileNetV2 network.First,the Block module in the MobileNetV2 structure is improved by adding SE blocks and scale attention mechanism,and a fully connected layer and global pooling is introduced to facilitate the acquisition of multi-scale features.Then,for the characteristics of large similarity between COVID-19 images and CAP images,GroupNorm is used instead of BatchNorm2d to enable the model to better acquire pneumonia features.Finally,the model is optimized using SGD optimizer.The experimental results show that the classification accuracy of the model proposed in this paper has a higher classification accuracy.

关 键 词：新冠病毒感染 计算机断层扫描图像 卷积神经网络 MobileNetV2-SELN 分组归一化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]