A probabilistic framework is presented that enables image registration, tissue classification, and bias correction to be combined within the same generative model. A derivation of a log-likelihood objective function for the unified model is provided. The model is based on a mixture of Gaussians and is extended to incorporate a smooth intensity variation and nonlinear registration with tissue probability maps. A strategy for optimising the model parameters is described, along with the requisite partial derivatives of the objective function.

The key challenge faced by a neurosurgeon is the removal from the brain of as much tumor tissue as possible while minimizing the removal of healthy tissue and avoiding the disruption of critical anatomical structures. We developed an algorithm to create enhanced visualizations of tumor and critical brain structures by aligning preoperatively acquired image data with intraoperative images of the patient's brain during surgery.

The accuracy of image-guided neurosurgery generally suffers from brain deformations due to intraoperative changes, e.g., brain shift or tumor resection. In order to improve the accuracy, we developed a biomechanical model of the human head which can be employed for the correction of preoperative images. By now, the model comprises two different materials. The correction of the preoperative image is driven by a set of given landmark correspondences. Our approach has been tested using synthetic images and yields physically plausible results. Additionally, we carried out registration experiments with a preoperative MR image and a corresponding postoperative image simulating an intraoperative image. We found, that our approach yields good prediction results, even in the case when correspondences are given in a small area of the image only.

omical templates with specific datasets, thus facilitating segmentation (i.e. segmentation by registration). More recently, these techniques have been used to capture changes which occur during neurosurgery. With the ongoing development of robust algorithms and advanced hardware platforms, further applications in surgical visualization and enhanced functional image analysis are inevitable. One exciting application of nonrigid registration algorithms is in the automatic registration of multimodal image data. Rigid registration of multimodal data has been greatly facilitated by the framework provided by mutual information (MI). However, MI-based strategies to effectively capture large nonrigid shape differences are still being explored. An alternate approach is to normalize multimodality images and thus reduce the problem to a monomodality match. In the first section, we present a nonrigid registration method which uses an intensity transform which allows a single intensity in one modal

In order to improve the accuracy of image-guided neurosurgery, dierent biomechanical models have been developed to correct preoperative images w.r.t. intraoperative changes like brain shift or tumor resection. All existing biomechanical models simulate dierent anatomical structures by using either appropriate boundary conditions or by spatially varying material parameter values, while assuming the same physical model for all anatomical structures. In general, this leads to physically implausible results, especially in the case of adjacent elastic and uid structures. Therefore, we propose a new approach which allows to couple dierent physical models. In our case, we simulate rigid, elastic, and uid regions by using the appropriate physical description for each material, namely either the Navier equation or the Stokes equation. To solve the resulting dierential equations, we derive a linear matrix system for each region by applying the nite element method (FEM). Thereafter, the li...

The estimation of soft tissue deformation from 3D image sequences is an important problem in a number of fields such as diagnosis of heart disease and image guided surgery. In this paper we describe a methodology for bridging biomechanical information regarding material properties with a Bayesian framework which allows for proper modeling of image noise in order to estimate these deformations. The resulting partial di#erential equations are discretized and solved using the finite element method. We demonstrate the application of this method to estimating strains from sequences of invivo left ventricular MR images, where we incorporate information about the fibrous structure of the ventricle. The deformation estimates obtained exhibit similar patterns with measurements obtained from Magnetic Resonance Tagging. An earlier version of this work will appear in [14]. Keywords: Non-rigid motion analysis, cardiac motion, continuum mechanics, soft tissue deformation. 1 1 Introduct...

The accuracy of image-guided neurosurgery generally suffers from brain deformations due to intraoperative changes, e.g. brain shift or tumor resection. In order to improve the accuracy, we developed a biomechanical model of the human head which can be employed for the correction of preoperative images. By now, the model comprises two different materials and the correction of the preoperative image is driven by a set of given image correspondences. Our approach has been tested on synthetic images and yields physically plausible results. Additionally, we carried out registration experiments with a preoperative MR image and a corresponding postoperative image. We found, that our approach yields good prediction results, even in the case when correspondences are given in a small area of the image only.

The non-invasive quantitative estimation of regional cardiac deformation has important clinical implications for the assessment of viability in the heart wall. In this work we describe a general framework for estimating soft tissue deformation from sequences of three-dimensional medical images. We also explore some of their theoretical constraints which can be used to guide the selection of an appropriate model for the displacement field. We then apply this framework to the problem of estimating left ventricular deformations from sequences of 3D image sequences. The images are segmented interactively to extract the endocardial and epicardial surfaces. Then, initial frame-to-frame correspondences are established between points on the surfaces using a shape-tracking approach. The myocardium is modeled using a transversely isotropic linear elastic model, which accounts for the preferential stiffness of the left ventricular myocardium along its fiber directions. The measurements and the model are integrated within a Bayesian estimation framework. The resulting equations are solved using the finite element method, to produce a dense displacement field for the whole of the left ventricle. The dense displacement field is, in turn, used to calculate the deformation of the heart wall in terms of the strains.