Probabilistic non-rigid registration of prostate images: Modeling and quantifying uncertainty

Abstract

Registration of pre- to intra-procedural prostate images needs to handle the large changes in position and shape of the prostate caused by varying rectal filling and patient positioning. We describe a probabilistic method for non-rigid registration of prostate images which can quantify the most probable deformation as well as the uncertainty of the estimated deformation. The method is based on a biomechanical Finite Element model which treats the prostate as an elastic material. We use a Markov Chain Monte Carlo sampler to draw deformation configurations from the posterior distribution. In practice, we simultaneously estimate the boundary conditions (surface displacements) and the internal deformations of our biomechanical model. The proposed method was validated on a clinical MRI dataset with registration results comparable to previously published methods, but with the added benefit of also providing uncertainty estimates which may be important to take into account during prostate biopsy and brachytherapy procedures.

A survey of prostate modeling for image analysis

“…However, biomechanics must consider the laws governing the mechanical parameters of the tissue (e.g., stiffness, elasticity, and compressibility), and in vivo or ex vivo experimental procedures are required to estimate these parameters. In most studies, these parameters are determined from one or more experiments [78] or from numerical simulations where the parameters are the variables to be adjusted [29,79,80].…”

“…Registration and segmentation [116] Finite element modeling Registration [31] Registration [80,134,135,138,139] Modeling and simulation [79] Simulation and predictive modeling of prostate motion [136] Modeling prostate motion [29] Reconstruction and animation of deformable volumetric objects [131] Tracking [137] reconstruction; meshing; and the integration of material properties (often Young's modulus and Poisson's ratio) and boundary conditions, such as any rigid constraint imposed by the pelvic bone and displacement of the rectal wall [79]. The first step concerns the segmentation (manual, semi-automatic or automatic) of the anatomical structure in the images (usually CT, MRI or ultrasound).…”

Learning statistical correlation for fast prostate registration in image‐guided radiotherapy

“…[10][11][12][13] So far, many prostate registration algorithms have been developed in the literature, which can be broadly classified into two categories. The first category of methods is mainly based on correspondence detection or interpolation, i.e., first detecting correspondences through the segmentation of corresponding organs and then interpolating the dense correspondences for the rest regions of the image using thin plate splines (TPSs) based interpolation methods, [14][15][16] finite element methods, [17][18][19][20] or other techniques. 20,21 For instance, Venugopal et al 15 used TPS to estimate the prostate motion given homologous landmark points in the two prostate images.…”

“…The first category of methods is mainly based on correspondence detection or interpolation, i.e., first detecting correspondences through the segmentation of corresponding organs and then interpolating the dense correspondences for the rest regions of the image using thin plate splines (TPSs) based interpolation methods, [14][15][16] finite element methods, [17][18][19][20] or other techniques. 20,21 For instance, Venugopal et al 15 used TPS to estimate the prostate motion given homologous landmark points in the two prostate images. Bharatha et al 19 used an elastic finite element model to align the preprocedural images with the intraprocedural images of the prostate and showed a significant increase in overlap between the registered preprocedural and intraprocedural prostate images, comparing to only using rigid transformation.…”

“…Bharatha et al 19 used an elastic finite element model to align the preprocedural images with the intraprocedural images of the prostate and showed a significant increase in overlap between the registered preprocedural and intraprocedural prostate images, comparing to only using rigid transformation. Risholm et al 20 described a probabilistic method for nonrigid registration of prostate images, which could quantify the most probable deformation as well as the uncertainty of the estimated deformation based on a biomechanical finite element model. Note that, although this category of prostate registration methods requires the detection of correspondences (which often involves manual selection or contouring), it has the advantage of ensuring the alignment of important features 22 such as the regions around the prostate boundary.…”