Deep learning is an important branch of a broader family of machine learning methods which attempt to learn useful features from data by using a computational model with multilayer neural networks. A deep learning network can form more abstract high-layer features through combination of the lower layer features to find the complex intrinsic characteristics of the data. Convolutional neural network is one of most researched deep networks because of its excellent performance in processing images. This paper firstly introduced the structural characteristics and training methods of deep learning networks, followed by the convolutional neural network, and finally discussed the application and development trend of deep learning network in the field of neuroimaging.