文章摘要

文章信息/Info

Title:: Pneumonia X-ray image classification by EfficientNet with convolutional block attention module

Author(s):: HUANG Min; MA Shuya; （College of Biomedical Engineering, South-Central Minzu University, Wuhan 430074, China）

Keywords:: Pneumonia; Image classification; EfficientNet; CBAM; Transfer learning

摘要:: 为提高临床上对肺炎X射线图像诊断的效率及准确率，本研究基于EfficientNet网络模型，融合卷积注意力模块(convolutional block attention module，CBAM）提出了一种识别肺炎和正常图像的分类算法。首先，对数据进行增强以防止过拟合现象；其次，通过CBAM模块提升网络对肺炎病灶区的特征提取能力；最后，使用迁移学习加速网络训练，以提升分类性能。结果表明，该算法分类准确率、召回率、AUC分别达98.29%、98.03%、99.69%，可辅助医生高效、准确地实现肺炎诊断。

Abstract:: In order to improve the efficiency and accuracy of X-ray diagnosis of pneumonia in clinic, We proposed a classification algorithm based on the EfficientNet model and integrate convolutional block attention module (CBAM) to recognize the pneumonia and normal images. Firstly, the data was enhanced to prevent overfitting. Secondly, CBAM was used to improve the feature extraction ability of pneumonia focus areas. Finally, transfer learning was used to accelerate network training to improve the classification performance. The results showed that the classification accuracy, recall rate and AUC of this method were 98.29%, 98.03%, 99.69%, respectively.It can assist doctors to realize the diagnosis of pneumonia efficiently and accurately.?