Registration techniques for the analysis of the brain shift in neurosurgery

doi:10.1016/S0097-8493(00)00034-0

Computers & Graphics

Volume 24, Issue 3, June 2000, Pages 385-389

https://doi.org/10.1016/S0097-8493(00)00034-0 Get rights and content

Abstract

The brain shift is a phenomenon that occurs during surgical operations on the opened head. It is a deformation of the brain which prohibits exact navigation with pre-operatively acquired tomographic scans since correlation between the image data and the actual anatomical situation invalidates quickly after opening the skull. In order to analyze the brain shift nonlinear registration of two data sets is performed. Thereby, one data set is obtained before and the other during the operation with an open magnetic resonance scanner. Using registration based on deformable surfaces, models of the pre- and the intra-operative brain are obtained. After efficient distance calculation color encoding of the models gives quantitative information. For further anatomical orientation these models are integrated into a representation of the data produced with direct volume rendering. Additionally, we suggest a voxel-based approach based on maximizing mutual information. This accounts for deformations of deeper lying structures considering the volume. Adaptively subdividing the data into piecewise linear patches and using 3D texture mapping, fast evaluation of the non-linear deformation is achieved

Introduction

The comprehensive diagnosis of diseases and lesions is considerably assisted by different tomographic imaging modalities like computed tomography (CT) and magnetic resonance imaging (MRI), since they provide various information and improve the spatial understanding of anatomical structures. Integrating such data into an operation with a target in the center of the brain makes the access easier and tremendously reduces the risk of hitting critical structures. Due to the rigid behavior of the skull it is possible to define a reliable transformation between the image data and the head of the patient in the beginning of an operation using a neuro-navigation system. Thereby, showing the position of an instrument in relation to the image data, it is intended to predict structures which are approached. However, depending on the drainage of cerebrospinal fluid and the movement or removal of tissue, the initial shape of the brain changes and leads to the brain shift phenomenon which results in great inaccuracies during the ongoing course of the operation [1]. Therefore, it is important to understand the correlation of all effects and to correct the shift of the brain. Data sets showing the head of the patient before and after the shift of the brain has occurred represent an important prerequisite of the analysis. Currently, they are mainly obtained with magnetic resonance scanners [2] or to a more limited range with ultrasound devices [3].

In order to obtain a more comprehensive insight into the brain shift phenomenon two registration approaches were implemented. As presented in Section 2, deformable surfaces are used to generate models of the brain surface in the pre- and the intra-operative stage. Color encoding after efficient distance calculation further enhances the visual impression. Since the patient is typically positioned differently for every scan initial alignment of the image data is required. Thereby, the registration procedure and the evaluation of the brain shift are combined. This is true since it is not possible to compute an appropriate mapping function without knowing which part of the data is influenced by the deformation. In order to account for the deformation of the whole volume, a voxel based registration approach is suggested in Section 3. It is based on mutual information and uses 3D texture mapping in combination with piecewise linear transformations for acceleration. Finally, visualization results are presented in Section 4 which demonstrate the value of our approach in clinical applications.

Section snippets

Deformable surface approach

The overall work-flow of the deformable surface approach consists of four steps: surface generation, surface registration, distance computing and visualization. At the beginning the surfaces of the pre- and the intra-operative brain are extracted. This is achieved with the approach of deformable surfaces presented by Lürig [4]. As an advantage, this method does not require a certain surface topology in contrast to the technique of Terzopoulus [5]. The actual surface generation consists of an

Voxel-based approach

With respect to clinical application, it is also of importance to evaluate the shift of structures which lie below the brain surface. Therefore, a voxel-based approach is used which allows to calculate volume deformations. In this context we aimed at using graphics hardware to significantly accelerate the calculation of the required non-linear transformation. For this purpose the transformed data set is divided into a set of 3D linear patches. They are efficiently sliced by exploiting the

Results

The presented surface-based registration approaches were applied to 12 different pairs of data sets which were obtained before and after the shift of the brain has occurred. This was achieved by an installation of a magnetic resonance scanner (Magnetom Open, Siemens) in the neurosurgical operating theater which allows intra-operative scanning of the patient with the opened head. All parameters were chosen identical for both scans.

As presented in Fig. 1 the visualization of color encoded

Conclusion and future work

The presented registration techniques emphasize the problem of brain shift during neurosurgical procedures which may result in great inaccuracies during neurosurgery. The visualization of the results clearly demonstrates this phenomenon by fusing surface and volume representations of the pre- and the intra-operative brain. This allows for a global analysis based on color encoding of geometric models and a local inspection by measuring the distance at a specific position. In order to achieve a

References (11)

Hill D, Maurer C, Wang M, Maciunas R, Barwise J, Fitzpatrick J. Estimation of intraoperative brain surface movement....
Fahlbusch R, Nimsky C, Ganslandt O, Steinmerier R, Buchfelder M, Huk W. The Erlangen concept of image guided surgery...
Bucholz R, Yeh D, Trobaugh J, McDurmont L, Sturm C, Baumann C, et al. The correction of stereotactic inaccuracy caused...
Lürig C, Kobbelt L, Ertl T. Deformable surfaces for feature based indirect volume rendering. Procedings of computer...
Terzopoulos D. Regularization of inverse visual problems involving discontinuities. IEEE Transactions on Pattern...

There are more references available in the full text version of this article.

Cited by (22)

Intraoperative acquisition of fMRI and DTI
2011, Neurosurgery Clinics of North America
Citation Excerpt :
Navigation systems have a decreasing accuracy during the course of a surgical procedure because these systems are normally based on preoperative data only. This decreasing accuracy is because of brain shift, that is, the intraoperative deformation of the brain, which is caused by tumor removal, brain swelling, the use of brain retractors, and cerebrospinal fluid drainage.39,40 Intraoperative imaging offers a possibility to compensate for the effects of brain shift.
Correction of susceptibility artifacts in diffusion tensor data using non-linear registration
2007, Medical Image Analysis
Citation Excerpt :
These results are summarized in Table 3. This paper presents an approach for non-linear registration applied for DTI distortion correction extending our previous work (Hastreiter and Ertl, 1998; Hastreiter et al., 2000; Soza et al., 2002; Hastreiter et al., 2004; Soza et al., 2004). In this context, the optimization process was considerably improved using the SPSA optimization strategy.
Diffusion tensor imaging can be used to localize major white matter tracts within the human brain. For surgery of tumors near eloquent brain areas such as the pyramidal tract this information is of importance to achieve an optimal resection while avoiding post-operative neurological deficits. However, due to the small bandwidth of echo planar imaging, diffusion tensor images suffer from susceptibility artifacts resulting in positional shifts and distortion. As a consequence, the fiber tracts computed from echo planar imaging data are spatially distorted. We present an approach based on non-linear registration using Bézier functions to efficiently correct distortions due to susceptibility artifacts. The approach makes extensive use of graphics hardware to accelerate the non-linear registration procedure. An improvement presented in this paper is a more robust and efficient optimization strategy based on simultaneous perturbation stochastic approximation (SPSA).
Since the accuracy of non-linear registration is crucial for the value of the presented correction method, two techniques were applied in order to prove the quality of the proposed framework. First, the registration accuracy was evaluated by recovering a known transformation with non-linear registration. Second, landmark-based evaluation of the registration method for anatomical and diffusion tensor data was performed. The registration was then applied to patients with lesions adjacent to the pyramidal tract in order to compensate for susceptibility artifacts. The effect of the correction on the pyramidal tract was then quantified by measuring the position of the tract before and after registration. As a result, the distortions observed in phase encoding direction were most prominent at the cortex and the brainstem. The presented approach allows correcting fiber tract distortions which is an important prerequisite when tractography data are integrated into a stereotactic setup for intra-operative guidance.
A study on medical image registration by mutual information with pyramid data structure
2007, Computers in Biology and Medicine
Citation Excerpt :
Medical image registration is a key technique in image processing and visualization, and it arises and develops with the expanded applications of medical imaging techniques. The registration goal is to combine the information in different modality images or to get rid of the artifacts in images, and it is now an indispensable technique for disease diagnosis and neuroscience research [15]. There are diverse medical image registration methods, including the traditional methods such as landmark based registration [1,2,14,19,23], feature based registration [3,22], and some relatively new information based methods such as the one based on mutual information (MI) [4,5,17,20].
The registration method based on mutual information is currently a popular technique for medical image registration, but the computation of the mutual information is complex and the registration speed is slow. In this work, a new slice accumulation pyramid (SAP) data structure was proposed to expedite the registration process. A numerical comparative study between the new data structure and the existing wavelet pyramid (WP) data structure was given, and the results confirmed that the new pyramid data structure was superior to the WP in both the calculation efficiency and the optimizing performance. Finally, SAP was applied to remove the artifacts between CT and MRI data sets, and the results showed the validation of SAP to registration of mulmodality images.
Quasi-real-time neurosurgery support by MRI processing via grid computing
2005, Neurosurgery Clinics of North America
Citation Excerpt :
Unfortunately, it is limited to anatomic data only and is also costly. Hastreiter et al [4] try to visualize brain shift using a deformable surface model, which they apply to pre- and intraoperative data sets to show the movement of the brain's surface. To take into account the shift of structures that lie below the brain surface, a voxel-based approach is used, which allows calculation of volume deformation.
Strategies for brain shift evaluation
2004, Medical Image Analysis
Citation Excerpt :
It represents an extension to pure software based approaches (Wells et al., 1995; Collignon et al., 1995; Studholme et al., 1996, 1999). Contrary to initial results presented in (Hastreiter et al., 2000), the approach became more robust using pre-segmented brain volumes. Additionally, the efficiency was further improved by the use of adaptive refinement of the underlying control point grid.
For the analysis of the brain shift phenomenon different strategies were applied. In 32 glioma cases pre- and intraoperative MR datasets were acquired in order to evaluate the maximum displacement of the brain surface and the deep tumor margin. After rigid registration using the software of the neuronavigation system, a direct comparison was made with 2D- and 3D visualizations. As a result, a great variability of the brain shift was observed ranging up to 24 mm for cortical displacement and exceeding 3 mm for the deep tumor margin in 66% of all cases. Following intraoperative imaging the neuronavigation system was updated in eight cases providing reliable guidance. For a more comprehensive analysis a voxel-based nonlinear registration was applied. Aiming at improved speed of alignment we performed all interpolation operations with 3D texture mapping based on OpenGL functions supported in graphics hardware. Further acceleration was achieved with an adaptive refinement of the underlying control point grid focusing on the main deformation areas. For a quick overview the registered datasets were evaluated with different 3D visualization approaches. Finally, the results were compared to the initial measurements contributing to a better understanding of the brain shift phenomenon. Overall, the experiments clearly demonstrate that deformations of the brain surface and deeper brain structures are uncorrelated.
Serial registration of intraoperative MR images of the brain
2002, Medical Image Analysis
Citation Excerpt :
Previous experience using multi-modal image similarity criteria (i.e. mutual information (MI), see (Hata et al., 1998; Hata, 1998)) has also shown that this cannot fully solve the contrast variation problems either. Hastreiter et al. (2000) propose to use an MI-based matching algorithm, but the images need to be acquired under very similar conditions for the algorithm to work. For improved visualization of the brain shift, they use a deformable surface model.
The increased use of image-guided surgery systems during neurosurgery has brought to prominence the inaccuracies of conventional intraoperative navigation systems caused by shape changes such as those due to brain shift. We propose a method to track the deformation of the brain and update preoperative images using intraoperative MR images acquired at different crucial time points during surgery. We use a deformable surface matching algorithm to capture the deformation of boundaries of key structures (cortical surface, ventricles and tumor) throughout the neurosurgical procedure, and a linear finite element elastic model to infer a volumetric deformation. The boundary data are extracted from intraoperative MR images using a real-time intraoperative segmentation algorithm. The algorithm has been applied to a sequence of intraoperative MR images of the brain exhibiting brain shift and tumor resection. Our results characterize the brain shift after opening of the dura and at the different stages of tumor resection, and brain swelling afterwards. Analysis of the average deformation capture was assessed by comparing landmarks identified manually and the results indicate an accuracy of 0.7±0.6 mm (mean±S.D.) for boundary surface landmarks, of 0.9±0.6 mm for landmarks inside the boundary surfaces, and 1.6±0.9 mm for landmarks in the vicinity of the tumor.

View all citing articles on Scopus

View full text

Data VisualizationRegistration techniques for the analysis of the brain shift in neurosurgery