【摘要】 当前肺结核严重威胁人们的生命健康，快速准确地诊断肺结核一直以来是常规影像学研究的重点与难点。由于医学图像的特殊性，往往难以获取充足的数据来进行深度学习训练。对此，提出一种适用于小数据集的肺结核X-ray图像分类模型。首先结合图像缩放、图像增强和目标区域提取等多种方法对原始X-ray图像进行预处理，在经过预处理后进行特征提取，通过在VGG-16网络中添加SE-Block来给图像的不同通道进行权重分配，并使用SVM对正常X-ray与肺结核X-ray进行分类。实验结果表明，所提模型的分类准确率与基线模型和其他已有模型对比都有显著提升，在小数据集中具有较好的分类性能。更多还原

【Abstract】 At present,pulmonary tuberculosis is a serious threat to people’s life and health. Rapid and accurate diagnosis of pulmonary tuberculosis has always been the focus and difficulty of conventional imaging studies. Due to the particularity of medical images,it is often difficult to obtain sufficient data for deep learning training. In this paper,a pulmonary tuberculosis X-ray image classification model suitable for small data sets is proposed. Firstly,image scaling,image enhancement and target region extraction are combined to preprocess the original X-ray image. After the preprocessing,feature extraction is carried out. SE-Block is added to VGG-16 network to assign weights to different channels of the image. SVM is used to classify normal X-ray and pulmonary tuberculosis X-ray. Experimental results show that the classification accuracy of the proposed model is significantly improved compared with the baseline model and other existing models,and has better classification performance in small data sets.更多还原