Deformable Medical Image Registration: A Survey

Abstract

Deformable image registration is a fundamental task in medical image processing. Among its most important applications, one may cite: i) multi-modality fusion, where information acquired by different imaging devices or protocols is fused to facilitate diagnosis and treatment planning; ii) longitudinal studies, where temporal structural or anatomical changes are investigated; and iii) population modeling and statistical atlases used to study normal anatomical variability. In this paper, we attempt to give an overview of deformable registration methods, putting emphasis on the most recent advances in the domain. Additional emphasis has been given to techniques applied to medical images. In order to study image registration methods in depth, their main components are identified and studied independently. The most recent techniques are presented in a systematic fashion. The contribution of this paper is to provide an extensive account of registration techniques in a systematic manner.

A Variational Formulation for Discrete Registration

“…Over the years, numerous regularization approaches have been proposed for deformable registration (see [2] for a detailed review). These approaches can be broadly categorized into two groups, namely non-parametric and parametric approaches.…”

Accuracy of eight deformable image registration (DIR) methods for tomotherapy megavoltage computed tomography (MVCT) images

“…Moreover, the results of the registration naturally depend on the deformation model. The choice of deformation algorithms and transformation frameworks in MVCT images is of great importance in the registration process, as that entails an important compromise between computational efficiency and richness of description 8. The popular tool‐kit for image registration, Deformable Image Registration and Adaptive Radiotherapy (DIRART) is a software suite for DIR plus ART.…”

“…The statistical analysis that follows registration is biased on the choice of the target domain, whereas symmetric transformation simultaneously estimates both the forward and backward transformations. The data matching term quantifies how well the images are aligned when one image is deformed by the forward transformation and the other image is deformed by the backward transformation 8. For the deformation algorithms, Original Demons and Original Horn & Schunck Optical Flow are non‐parametric deformation algorithms based on a vector per voxel method that describes the displacement to model the deformation of the anatomy for well‐studied models of fluid flow or the deformation of a viscoelastic material 11.…”

“…Regarding the mapping direction, when the forward mapping is estimated, every voxel of the source image is pushed forward to its estimated position in the target image. When the backward mapping is estimated, the pixel value of a voxel in the target image is pulled from the source image 8. Therefore, the sequences of the target domain were the effect of the deformable registration process.…”

A quadrature filter approach for registration accuracy assessment of fundus images

“…Depending on the task at hand one chooses a specific registration method. The state-of-the-art medical image registration techniques are presented in recent surveys [1,2]. In addition to robustness, a key performance measure of these techniques is accuracy.…”