基于空间转换网络的肺结节分类方法  

Classification of pulmonary nodules based on spatial transformation network

作　　者：赵赫威 王华军[1] 卞华军 ZHAO He-wei;WANG Hua-jun;BIAN Hua-jun(School of Computer and Network Security,Chengdu University of Technology,Chengdu 610000,China)

机构地区：[1]成都理工大学计算机与网络安全学院,成都610000

出　　处：《信息技术》2023年第10期6-11,16,共7页Information Technology

基　　金：四川省教育厅自然科学基金项目(2020RYJ02,MSSB-2020-10)。

摘　　要：为了降低肺结节分类问题中因样本的形变、扭曲、偏移等对分类结果的影响,提出一种基于空间转换网络的特征提取网络模型。在该模型中,利用空间转换网络的自适应仿射变换能力,赋予分类网络以自主矫正样本的能力,提升分类模型的鲁棒性。经过实验证明,该方法能够有效提高分类任务的准确率。In order to reduce the influence of sample deformation,distortion and offset on the classification results in the classification of pulmonary nodules,a feature extraction network model based on spatial transformation network is proposed.In this model,the adaptive affine transformation ability of spatial transformation network is used to endow the classification network with the ability of self-correcting samples,which improves the robustness of the classification model.A large number of experiments show that this method can effectively improve the accuracy of classification tasks.

关 键 词：肺结节 良恶性分类 空间转换网络 深度神经网络 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]