Pulmonary nodule benign and malignant identification method based on attention mechanism

The invention relates to a pulmonary nodule benign and malignant identification method, in particular to the pulmonary nodule benign and malignant identification method based on the attention mechanism. According to the method, a 3D Attention DPN network is used as a feature extraction network, so that the network pays more attention to detail features of a nodule region in channel and space. According to the classification network, the advantages of high feature extraction speed and high accuracy of a DPN network are reserved, the capacity of a CBAM attention module for conducting self-adaptive adjustment on features from the channel dimension and the space dimension is integrated, and benign and malignant recognition can be effectively conducted on nodules in combination with an advancedclassifier (GBM). 本发明涉及一种肺结节良恶性识别方法，特别是一种基于注意力机制的肺结节良恶性识别方法。该方法将3D Attention DPN网络作为特征提取网络，让网络从通道和空间上更加注重结节区域的细节特征。在本发明的分类网络中既保留了DPN网络对特征提取速度快、准确度高的优势，又集成了CBAM（Convolutional Attention Block Module）注意力模块从通道和空间两个维度对特征进行自适应调整的能