Bayesian estimation of intra-operative deformation for image-guided surgery using 3-D ultrasound

Abstract

This paper describes the application of Bayesian theory to the problem of compensating for soft tissue deformation to improve the accuracy of image-guided surgery. A triangular surface mesh segmented from a pre-operative image is used as the input to the algorithm, and intra-operatively acquired ultrasound data compounded into a 3-D volume is used to guide the deformation process. Prior probabilities are defined for the boundary points of the segmented structure based on knowledge of the direction of gravity, the position of the surface of the surgical scene, and knowledge of the tissue properties. The posterior probabilities of the locations of each of the boundary points are then maximised according to Bayes’ theorem. A regularisation term is included to constrain deformation to the global structure of the object. The technique is demonstrated using a deformable phantom designed to have similar properties to human tissue. Results presented demonstrate that the algorithm was able to recover much of the deformation for a number of objects at varying depths from the source of deformation. This technique offers a convenient means of introducing prior knowledge of the operative situation into the problem of soft tissue deformation and has the potential for greatly improving the utility of image-guided surgery.