Oskar M. Skrinjar

Functional MRI of Language Processing: Dependence on Input Modality and Temporal Lobe Epilepsy

Summary: u2002Purpose: Functional magnetic resonance imaging (MRI) using two language‐comprehension tasks was evaluated to determine its ability to lateralize language processing and identify regions that must be spared in surgery.

Model-driven brain shift compensation

Surgical navigation systems provide the surgeon with a display of preoperative and intraoperative data in the same coordinate system. However, the systems currently in use in neurosurgery are subject to inaccuracy caused by intraoperative brain deformation (brain shift), since they typically assume that the intracranial structures are rigid. Experiments show brain shift of up to 1 cm, making it the dominant error in the system. We propose a biomechanical-model-based approach for brain shift compensation. Two models are presented: a damped spring-mass model and a model based on continuum mechanics. Both models are guided by limited intraoperative (exposed brain) surface data, with the aim to recover the deformation in the full volume. The two models are compared and their advantages and disadvantages discussed. A partial validation using intraoperative MR image sequences indicates that the approach reduces the error caused by brain shift.

Real Time 3D Brain Shift Compensation

Surgical navigation systems are used intraoperatively to provide the surgeon with a display of preoperative and intraoperative data in the same coordinate system and help her or him guide the surgery. However, these systems are subject to inaccuracy caused by intraoperative brain movement (brain shift) since commercial systems in use today typically assume that the intracranial structures are rigid. Experiments show brain shifts up to several millimeters, making it the cause of the dominant error in the system. We propose an image-based brain shift compensation system based on an intraoperatively guided deformable model. We have recorded a set of brain surface points during the surgery and used them to guide and/or validate the model predictions. Initial results show that this system limits the error between its brain surface prediction and real brain surfaces to within 0.5 mm, which is a significant improvement over the systems that are based on the rigid brain assumption, that in this case would have an error of 3 mm or greater.

Brain Shift Modeling for Use in Neurosurgery

Surgical navigation systems are used intraoperatively to help the surgeon to ascertain her or his position and to guide tools within the patient frame with respect to registered structures of interest in the preoperative images. However, these systems are subject to inaccuracy caused by intraoperative brain movement (brain shift) since they assume that the intracranial structures are rigid. Experiments show brain shifts of up to several millimeters, making it the cause of the dominant error in those systems. We propose a method for reducing this error based on a dynamic brain model. The initial model state is obtained from preoperative data. The brain tissue is modeled as a homogeneous linear visco-elastic material, although the model allows for setting the tissue properties locally. Gravity draws the brain downwards which in turn interacts with the skull and other surrounding structures. The simulation results are presented both for a 2D model (the mid-sagittal slice) and a 3D model. The results show the time evolution of the brain deformation. The complete 3D validation of the simulated brain deformation is a rather complicated task and is currently in progress within our laboratory, but a procedure is proposed for updating the model in time by one or more of several intraoperative measurements.

Steps Toward a Stereo-Camera-Guided Biomechanical Model for Brain Shift Compensation

Surgical navigation systems provide the surgeon with a display of preoperative and intraoperative data in the same coordinate system. However, the systems currently in use in neurosurgery are subject to inaccuracy caused by intraoperative brain movement (brain shift) since they typically assume that the intracranial structures are rigid. Experiments show brain shift of up to one centimeter, making it the dominant error in the system. We propose a system that compensates for this error. It is based on a continuum 3D biomechanical deformable brain model guided by intraoperative data. The model takes into account neuro-anatomical constraints and is able to correspondingly deform all preoperatively acquired data. The system was tested on two sets of intraoperative MR scans, and an initial validation indicated that our approach reduced the error caused by brain shift.

A stereo-guided biomechanical model for volumetric deformation analysis

A framework for soft tissue deformation recovery is presented and tested on both simulated and real data. The core of the approach is a volumetric deformable model based on the biomechanics of the material. Often, e.g. in medical applications, the deformation is very complex and dependent on many factors. In such cases, in order to reliably estimate the deformation, the model has to be guided by the available, typically limited, data. We propose the use of stereo cameras for guidance, where the reconstructed surface is used as a boundary condition for the partial differential equations that define the model. By integrating the model deformation and surface reconstruction algorithms, we avoid the correspondence problem and reduce the impact of surface specularities. The approach is applied to the problem of brain shift brain deformation that occurs during the surgery. A partial validation using intraoperative magnetic resonance images suggests that the error introduced by the brain shift can be reduced almost to the image resolution.

Surface growing from stereo images

We present a new theoretical result for the problem of surface reconstruction from stereo images. For a given initial seed point, i.e. for a pair of corresponding points in the left and right image, the proposed algorithm grows the surface without directly computing the point correspondences. The method assumes the Lambertian surface reflectance model. Our approach is based on a surface normal calculation from the left and right image gradients. Knowing the surface normal, the algorithm grows the surface in the directions defined by the tangent plane. The algorithm is independent of the camera model, and requires placement of an initial seed point for each surface to be reconstructed. Technical problems associated with errors in the image gradient estimates and camera calibration are discussed and a solution is suggested. In addition to this algorithm, we present a theoretical result that permits one to track surfaces deforming in time, which is often encountered in medical applications (e.g. brain surface deforms during the surgery). These methods of surface reconstruction and deformable surface tracking are applied to the particular problem of brain shift, commonly recognized as one of the main source of errors in surgical navigation systems used in neurosurgery. We also suggest a way to overcome problems associated with brain surface specularities caused by fluids on the brain surface and lights in the operating room. We conclude with experimental results on real brain images and show that the surface reconstruction algorithm is robust to the position of the initial seed point.

Automatic Extraction of Implanted Electrode Grids

It is common in epilepsy surgery to implant grids and strips of electrodes between the skull and brain or inside the brain, in order to localize functional areas. MR scans are currently used for a variety of image-guided surgical planning tasks, including the localization of the electrode grids. However, the MR scan taken of a patient with implanted electrodes is distorted, and it is difficult to visualize and relate the electrode positions to head and brain structures. For this reason we have developed an automatic algorithm that reliably extracts grids of electrodes from corrupted post-op MR scans. The grid is fitted as a smooth, curved surface through the estimated electrode positions, properly estimating the orientation of the thin disk-shaped electrodes. The extracted grid is then displayed in 3D together with the desired brain structures, coloring the electrodes corresponding to particular functional areas. It is now much easier to visualize and locate the positions of the important functional areas with respect to other brain structures and plan the surgery. This method is currently in clinical use within the Department of Neurosurgery, Yale University and Yale New Haven Hospital.