Deep learning-based methods have recently demonstrated promising results in\ndeformable image registration for a wide range of medical image analysis tasks.\nHowever, existing deep learning-based methods are usually limited to small\ndeformation settings, and desirable properties of the transformation including\nbijective mapping and topology preservation are often being ignored by these\napproaches. In this paper, we propose a deep Laplacian Pyramid Image\nRegistration Network, which can solve the image registration optimization\nproblem in a coarse-to-fine fashion within the space of diffeomorphic maps.\nExtensive quantitative and qualitative evaluations on two MR brain scan\ndatasets show that our method outperforms the existing methods by a significant\nmargin while maintaining desirable diffeomorphic properties and promising\nregistration speed.\n