基于注意力机制的肺结节分类研究

CLASSFICATION OF PULMONARY NODULES BASED ON ATTENTION MECHANISM

为进一步提升肺结节分类的效果,引入一种基于注意力机制的分类算法。通过在神经网络中添加空间和通道注意力因子,使得肺结节分类网络生成更有效的特征映射,结合梯度提升树算法,进一步提升模型的性能。经过大量实验后,证明了该方法的有效性。

In order to further improve the effect of lung nodule classification,a classification algorithm based on attention mechanism is introduced.By adding spatial and channel attention factors to the neural network,the lung nodule classification network generates more effective feature maps,which is combined with the gradient boosting tree algorithm to further improve the performance of the model.After a lot of experiments,the effectiveness of the this method was proved.