MDReg‐Net: Multi‐resolution diffeomorphic image registration using fully convolutional networks with deep self‐supervision

Abstract

We present a diffeomorphic image registration algorithm to learn spatial transformations between pairs of images to be registered using fully convolutional networks (FCNs) under a self-supervised learning setting. The network is trained to estimate diffeomorphic spatial transformations between pairs of images by maximizing an image-wise similarity metric between fixed and warped moving images, similar to conventional image registration algorithms. It is implemented in a multi-resolution image registration framework to optimize and learn spatial transformations at different image resolutions jointly and incrementally with deep self-supervision in order to better handle large deformation between images. A spatial Gaussian smoothing kernel is integrated with the FCNs to yield sufficiently smooth deformation fields to achieve diffeomorphic image registration. Particularly, spatial transformations learned at coarser resolutions are utilized to warp the moving image, which is subsequently used for learning incremental transformations at finer resolutions. This procedure proceeds recursively to the full image resolution and the accumulated transformations serve as the final transformation to warp the moving image at the finest resolution. Experimental results for registering high resolution 3D structural brain magnetic resonance (MR) images have demonstrated that image registration networks trained by our method obtain robust, diffeomorphic image registration results within seconds with improved accuracy compared with state-of-the-art image registration algorithms.

DeepRetroMoCo: Deep neural network-based Retrospective Motion Correction Algorithm for Spinal Cord functional MRI

“…We don't need this information, which is another clear distinction between our approach and earlier research. The previous two works (Li & Fan, 2017;Vos, Berendsen, Viergever, Staring, & Išgum, 2017) reported unsupervised methods that are close to ours. Both use the CNN neural network with spatial transformation function (Jaderberg et al, 2015), which warps images on top of each other and has significant problems: they only operate on a limited subset of volumes and only support small transformations.…”

Generative Adversarial Registration for Improved Conditional Deformable Templates

“…Instead, a series of registration algorithms more suited to biomedical images have been developed [23,30,50,83,92], further leading to several topology-preserving diffeomorphic extensions [8,12,20,96,98,105,89]. More recently, deep networks trained under either supervised [19,90,103] or unsuper-vised [11,26,29,57,60,73] registration objectives have emerged, simultaneously offering both greater modeling flexibility and accelerated inference performance.…”

Joint Unsupervised Infrared-RGB Video Registration and Fusion

“…Challenges associated with generating ground truth data have recently led many researchers to develop unsupervised frameworks. Two recent works [4,17], presents an unsupervised learning-based image registration methods. Both propose a neural network consisting of a CNN and a spatial transformation function [10] that warps images to one another.…”

“…Both propose a neural network consisting of a CNN and a spatial transformation function [10] that warps images to one another. However, these two initial methods are only demonstrated on limited subsets, such as 3D sub-regions [17] or 2D slices [4], and support only small transformations [4]. All above methods were demonstrated on medical images.…”