Thomas Hartkens

A Generic Framework for Non-rigid Registration Based on Non-uniform Multi-level Free-Form Deformations

This work presents a framework for non-rigid registration which extends and generalizes a previously developed technique by Rueckert et al. [1]. We combine multi-resolution optimization with free-form deformations (FFDs) based on multi-level B-splines to simulate a non-uniform control point distribution. We have applied this to a number of different medical registration tasks to demonstrate its wide applicability, including interventional MRI brain tissue deformation compensation, breathing motion compensation in liver MRI, intra-modality inter-modality registration of pre-operative brain MRI to CT electrode implant data, and inter-subject registration of brain MRI. Our results demonstrate that the new algorithm can successfully register images with an improved performance, while achieving a significant reduction in run-time.

Measurement and analysis of brain deformation during neurosurgery

Recent studies have shown that the surface of the brain is deformed by up to 20 mm after the skull is opened during neurosurgery, which could lead to substantial error in commercial image-guided surgery systems. We quantitatively analyze the intraoperative brain deformation of 24 subjects to investigate whether simple rules can describe or predict the deformation. Interventional magnetic resonance images acquired at the start and end of the procedure are registered nonrigidly to obtain deformation values throughout the brain. Deformation patterns are investigated quantitatively with respect to the location an magnitude of deformation, and to the distribution and principal direction of the displacements. We also measure the volume change of the lateral ventricles by manual segmentation. Our study indicates that brain shift occurs predominantly in the hemisphere ipsi-lateral to the craniotomy, and that there is more brain deformation during resection procedures than during biopsy or functional procedures. However, the brain deformation patterns are extremely complex in this group of subjects. This paper quantitatively demonstrates that brain deformation occurs not only at the surface, but also in deeper brain structure, and that the principal direction of displacement does not always correspond with the direction of gravity. Therefore, simple computational algorithms that utilize limited intraoperative information (e.g., brain surface shift) will not always accurately predict brain deformation at the lesion.

Longitudinal regional brain volume changes quantified in normal aging and Alzheimer's APP × PS1 mice using MRI

In humans, mutations of amyloid precursor protein (APP) and presenilins (PS) 1 and 2 are associated with amyloid deposition, brain structural change and cognitive decline, like in Alzheimers disease (AD). Mice expressing these proteins have illuminated neurodegenerative disease processes but, unlike in humans, quantitative imaging has been little used to systematically determine their effects, or those of normal aging, on brain structure in vivo. Accordingly, we investigated wildtype (WT) and TASTPM mice (expressing human APP(695(K595N, M596L)) x PS1(M146V)) longitudinally using MRI. Automated global and local image registration, allied to a standard digital atlas, provided pairwise segmentation of 13 brain regions. We found the mature mouse brain, unlike in humans, enlarges significantly from 6-14 months old (WT 3.8+/-1.7%, mean+/-SD, P<0.0001). Significant changes were also seen in other WT brain regions, providing an anatomical benchmark for comparing other mouse strains and models of brain disorder. In TASTPM, progressive amyloidosis and astrogliosis, detected immunohistochemically, reflected even larger whole brain changes (5.1+/-1.4%, P<0.0001, transgenexage interaction P=0.0311). Normalising regional volumes to whole brain measurements revealed significant, prolonged, WT-TASTPM volume differences, suggesting transgene effects establish at <6 months old of age in most regions. As in humans, gray matter-rich regions decline with age (e.g. thalamus, cerebral cortex and caudoputamen); ventricles and white matter (corpus callosum, corticospinal tract, fornix system) increase; in TASTPMs such trends often varied significantly from WT (especially hippocampus). The pervasive, age-related structural changes between WT and AD transgenic mice (and mouse and human) suggest subtle but fundamental species differences and AD transgene effects.

VTK CISG Registration Toolkit An Open Source Software Package for Affine and Non-rigid Registration of Single- and Multimodal 3D Images

Voxel-based image registration using Normalised Mutual Information (NMI) has been shown to register single- and multi-modal 3D images accurately without any user interaction [1,5,4]. Our group has proposed both an affine and a non-rigid registration algorithm based on NMI, and has validated these algorithms on a range of medical applications like brain, breast, and cardiac data [7, 8, 6, 2, 3]. We present a publicly available software package that incorporates these algorithms in a user-friendly command-line and graphical interface including a visualisation tool for 3D image pairs in order to analyse registration results. Beside the pure usage of registration algorithms, the software can be easily adjusted to specific environments (e.g. including other image file formats) and can be modified for specific applications.

Using Points and Surfaces to Improve Voxel-Based Non-rigid Registration

Voxel-based non-rigid registration algorithms have been successfully applied to a wide range of image types. However, in some cases the registration of quite different images, e.g. pre- and post-resection images, can fail because of a lack of voxel intensity correspondences. One solution is to introduce feature information into the voxel-based registration algorithms in order to incorporate higher level information about the expected deformation.We illustrate using one voxel-based registration algorithm that the incorporation of features yields considerable improvement of the registration results in such cases.

Constructing Patient Specific Models for Correcting Intraoperative Brain Deformation

In this work we present a Mesh Warping technique for the construction of patient-specific Finite Element Method models from patient MRI images, and demonstrate how simulated surgical loading can be applied to these models. We compare the results of this simulation with observed deformation during surgery, and show that our model matches well with the observed degree of deformation.

A Dynamic Brain Atlas

We describe a dynamic atlas that can be customized to an individual study subject in near-real-time. The atlas comprises 180 brain volumes each of which has been automatically segmented into grey matter, white matter and CSF, and also non-rigidly registered to the Montreal Brain Web reference dataset providing automatic delineation of brain structures of interest. To create a dynamic atlas, the user loads a study dataset (eg: a patient) and queries the atlas database to identify similar subjects. All selected database subjects are then aligned with the study subject using affine registration, and average tissue probability maps and structure delineations produced. The system can run on distributed data and distributed CPUs illustrating the potential of computational grids in medical image analysis.

Registration-based mesh construction technique for finite-element models of brains

The generation of patient specific meshes for Finite Element Methods (FEM) with application to brain deformation is a time consuming process, but is essential for the modeling of intra-operative deformation of the brain during neurosurgery using FEM techniques. We present an automatic method for the generation of FEM meshes fitting patient data. The method is based on non-rigid registration of patient MR images to an atlas brain image, followed by deformation of a high-quality mesh of this atlas brain. We demonstrate the technique on brain MRI images from 12 patients undergoing neurosurgery. We show that the FEM meshes generated by our technique are of good quality. We then demonstrate the utility of these FEM meshes by simulating simple neuro-surgical scenarios on example patients, and show that the deformations predicted by our brain model match the observed deformations. The meshes generated by our technique are of good quality, and are suitable for modeling the types of deformation observed during neurosurgery. The deformations predicted by a simple loading scenario match well with those observed following the actual surgery. This paper does not attempt an exhaustive study of brain deformation, but does provide an essential tool for such a study - a method of rapidly generating Finite Element Meshes fitting individual subject brains.

Deformation based morphometry analysis of serial magnetic resonance images of mouse brains

Deformation based morphometry is used to detect differences in in-vivo Magnetic Resonance Image (MRI) of the mouse brain obtained from two transgenic strains: TASTPM mice that over-express proteins associated with Alzheimers disease, and wild-type mice. MRI was carried out at four time points. We compare two different methods to detect group differences in the longitudinal and cross-sectional data. Both methods are based on non-rigid registration of the images to a mouse brain atlas. The whole brain volume measurements on 27 TASTPM and wild-type animals are reproducible to within 0.4% of whole brain volume. The agreement between different methods for measuring volumes in a serial study is shown. The ability to quantify changes in growth between strains in whole brain, hippocampus and cerebral cortex is demonstrated.

Intra-operative brain deformation using non-rigid image registration on a shared-memory multiprocessor computer

One major problem with non-rigid image registration techniques is their high computational cost. Because of this, these methods have found limited application to clinical situations where fast execution is required, e.g., intra-operative imaging. This paper applies a parallel implementation of a non-rigid image registration algorithm to pre and intra-operative MR images and quantitatively analyzes its scaling properties. The method computes the intra-operative brain deformation in about one minute using 64 CPUs on a 128-CPU shared-memory supercomputer (SGI Origin 3800). The serial component is no more than 2 percent of the total computation time, allowing a speedup of at least a factor of 50. In most cases, the theoretical limit of the speedup is substantially higher (up to 132-fold in the application examples presented in this paper). Our parallel algorithm is therefore capable of solving non-rigid registration problems with short execution time requirements and may be considered an important step in the application of such techniques to clinically important problems such as the computation of brain deformation during cranial image-guided surgery.