基于改进DenseNte的膝骨关节炎分类算法  

Knee osteoarthritis classification algorithm based on improved DenseNet network

作　　者：徐睿 刘爽 宋宇[1] 王昕[1] XU Rui;LIU Shuang;SONG Yu;WANG Xin(School of Computer Science&Engineering,Changchun University of Technology,Changchun 130102,China;Hunan Taoshang Medical Instrument Co.Ltd.,Changsha 410600,China)

机构地区：[1]长春工业大学计算机科学与工程学院,吉林长春130102 [2]湖南涛尚医疗器械有限公司,湖南长沙410600

出　　处：《长春工业大学学报》2024年第4期370-376,共7页Journal of Changchun University of Technology

基　　金：吉林省自然科学基金项目(20220101128JC)。

摘　　要：针对膝骨关节炎分类准确率低的问题,提出一种改进的DenseNet算法。在Dense201模型的基础上,加入浅层卷积块,将浅层特征与深层特征拼接,丰富分类层判别信息。利用水平翻转、图像旋转等数据增强方法扩增数据集,使用迁移学习对模型进行训练,减少过拟合。对比实验结果表明,改进后的DenseNet模型分类准确率达到91.0%,与原始DenseNet201网络相比提高了分类准确性。Aiming at the problem of low classification accuracy of knee osteoarthritis,an improved DenseNet algorithm is proposed.On the basis of the Dense201 model,a shallow convolution block is added,and the shallow features are spliced with the deep features to enrich the discriminant information of the classification layer.Use data enhancement methods such as horizontal flipping and image rotation to expand the data set,and use transfer learning to train the model to reduce overfitting.The results of comparative experiments show that the classification accuracy of the improved DenseNet model reaches 91.0%,which improves the classification accuracy compared with the original DenseNet network.

关 键 词：膝骨关节炎 磁共振图像 DenseNet 卷积神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]