基于密集神经网络的肺结节假阳性筛查模型

False Positive Screening of Pulmonary Nodules with DenseNet

通过肺部CT影像进行肺结节检测是肺癌早期筛查的重要手段,而候选结节的假阳性筛查是结节检测的关键部分。传统的结节检测方法主要通过简单的先验知识再利用低级的描述特征进行辅助检测,存在着假阳性高、敏感度低的问题。在深度学习中,卷积神经网络可以在通用的学习过程中提取图像的特征。提出了一种基于密集神经网络的结节假阳性筛查模型:首先对CT图像进行阈值分割提取肺区再截取以结节为中心的图像,送入网络模型进行分类训率;在网络模型中,通过稠密连接强化特征利用、扩大特征空间,采用瓶颈层降低参数冗余。模型在公开的LIDC数据集上取得了95.82%的准确率,ROC曲线下面积达到0.987,CPM为0.772。实验结果表明了该模型的有效性,其性能优于相关文献的方法,适用于肺结节的假阳性降低。

Pulmonary nodule detection through Computer Tomography images is the most primary way of early detection for lung cancer,and false positive screen is one of the most vital steps in automatic pulmonary nodule detection.Traditional pulmonary nodule detection methods mainly use simple prior knowledge and then use low-level description features for auxiliary detection.There are problems with high false positives and low sensitivities in the traditional methods.In deep learning,convolutional neural network can acquire image features in general learning process.We propose a model for reducing false positives of pulmonary nodules based on DenseNet.First,the threshold method is used to segment the lung area,and then the image centered on the nodule is intercepted and sent to the model.The model uses dense connection to achieve feature reuse and features reduction through the transfer layer.On 888 scans of the publicly available LIDC dataset,the model achieves high detection accuracy of 95.82%,receiver operating characteristic curve of 0.987,and the CPM is 0.772.The experiment demonstrates the effectiveness of the model,which is suitable for reducing false positives in pulmonary nodule detection systems.