Rupert Brooks

Generalizing Inverse Compositional and ESM Image Alignment

Inverse compositional (IC) image alignment (Baker and Matthews in Int. J. Comput. Vis. 56(3):221–255, 2004) uses the symmetry between the roles of the fixed and moving images for faster processing. However, it requires implementation of compositional optimizer update steps. The IC approach can be viewed as an efficient way of computing the similarity measure derivative relative to the fixed image warp parameters. Since the mapping between the fixed and moving warp parameters is continuous and differentiable, this derivative can be converted into the moving warp space using the chain rule. This avoids the need for compositional update steps. Our generalization also allows the efficient second order method (ESM) (Malis in Proceedings of the 2004 IEEE International Conference on Robotics and Automation (ICRA04), pp. 1843–1848, 2004; Benhimane and Malis in IEEE/RSJ International Conference on Intelligent Robots and Systems, 2004; Malis and Benhimane in Robot. Auton. Syst. 52(1):39–52, 2005) to be applied to general parameterizations of the transformation.Experiments using multiple similarity measures and optimizers show that our generalized IC method equals or exceeds the performance of the original IC approach. The generalized ESM approach is more reliable than the classic approach as it increases the capture radius of the optimization.

Deformable Ultrasound Registration without Reconstruction

Ultrasound (US) imaging is often proposed as an interoperative imaging modality. This use nearly always requires that the collected data be registered to preoperative data of another modality. Existing intensity-based registration approaches all begin by reconstructing a 3D US volume from the collected 2D slices. We propose to directly register the set of 2D slices to the preoperative images. We argue this has a number of advantages, including the omission of the potentially complex reconstruction step, greater adaptability of the similarity measures, and easier parallelization. We describe a system for performing this task and present results on phantom data that show that our slice based method consistently outperforms a reconstruction based method in both speed and accuracy.

A Computer Model of Soft Tissue Interaction with a Surgical Aspirator

Surgical aspirators are one of the most frequently used neurosurgical tools. Effective training on a neurosurgery simulator requires a visually and haptically realistic rendering of surgical aspiration. However, there is little published data on mechanical interaction between soft biological tissues and surgical aspirators. In this study an experimental setup for measuring tissue response is described and results on calf brain and a range of phantom materials are presented. Local graphical and haptic models are proposed. They are simple enough for real-time application, and closely match the observed tissue response. Tissue resection (cutting) with suction is simulated using a volume sculpting approach. A simulation of suction is presented as a demonstration of the effectiveness of the approach.

A fast discriminant approach to active object recognition and pose estimation

This paper presents a new criterion for viewpoint selection in the context of active Bayesian object recognition and pose estimation. Recognition is performed by probabilistically fusing successive observations with the current belief state of the system. Based on the current belief state, the next viewpoint is chosen to maximize the expected discriminability of the current competing hypotheses. Experiments on a difficult database of aircraft models show that this approach achieves comparable recognition performance to the widely used information theoretic approaches at a much lower computational cost.

Anytime similarity measures for faster alignment

Image alignment refers to finding the best transformation from a fixed reference image to a new image of a scene. This process is often optimizing a similarity measure between images, computed based on the image data. However, in time-critical applications state-of-the-art methods for computing similarity are too slow. Instead of using all the image data to compute similarity, one could use only a subset of pixels to improve the speed, but often this comes at the cost of reduced accuracy. These kinds of tradeoffs between the amount of computation and the accuracy of the result have been addressed in the field of real-time artificial intelligence as deliberation control problems. We propose that the optimization of a similarity measure is a natural application domain for deliberation control using the anytime algorithm framework. In this paper, we present anytime versions for the computation of two common image similarity measures: mean squared difference and mutual information. Off-line, we learn a performance profile specific to each measure, which is then used on-line to select the appropriate amount of pixels to process at each optimization step. When tested against existing techniques, our method achieves comparable quality and robustness with significantly less computation.

Virtual Reality Simulator: Demonstrated Use in Neurosurgical Oncology

Background. The overriding importance of patient safety, the complexity of surgical techniques, and the challenges associated with teaching surgical trainees in the operating room are all factors driving the need for innovative surgical simulation technologies. Technical development. Despite these issues, widespread use of virtual reality simulation technology in surgery has not been fully implemented, largely because of the technical complexities in developing clinically relevant and useful models. This article describes the successful use of the NeuroTouch neurosurgical simulator in the resection of a left frontal meningioma. Conclusion. The widespread application of surgical simulation technology has the potential to decrease surgical risk, improve operating room efficiency, and fundamentally change surgical training.

Haptic collision handling for simulation of transnasal surgery

Simulation of endoscopic navigation in the narrow nasal cavity poses important challenges to the computation of adequate and near‐realistic collision response and haptic feedback because extensive multidirectional contact and massive tissue deformations are inevitable. We present a virtual coupling algorithm that provides stable collision response as well as intuitive and smooth haptic interaction in all phases of the simulation. In each iteration, continuous collision detection between the point shell representing the surface of the virtual patient anatomy and the endoscope, represented by a cylinder, is performed. This allows for rolling back the instrument movement to the point in time the first collision occurred. Subsequently, a relaxation process locally optimizes the position and orientation of the instrument. A novel method of applying contact forces to colliding tissues and thus triggering appropriate deformations improves the fluency of navigation. This paper describes the algorithm and presents experimental results.

Generalizing inverse compositional image alignment

The inverse compositional (IC) approach to image alignment uses characteristics of the alignment problem to improve optimization speed. While a number of authors have noted its usefulness, to date it has only been explored for least-squares type image difference measures using Gauss-Newton optimization schemes. We extend the IC approach to general difference measures, and a wider class of optimization approaches, with specific development for normalized correlation and mutual information using the BFGS optimizer. We present alignment experiments on image pairs of several different classes that demonstrate performance improvements for the general case

The importance of scale when selecting pixels for image registration

Direct methods of image registration work by defining a measure of the difference between two images and using numerical optimization methods to find the transformation that minimizes the difference. It has often been proposed that these methods may be speeded up by using only a sub- set of pixels to compute the difference measure. Previous work has suggested some criteria to use in pixel selection based on the derivative of the image, but has not addressed the issue of performance degradation that can result from applying these techniques. In this paper, we show that un- less applied carefully, these methods do not actually help. Specifically, reliability of the registration algorithm is lost if the initial starting position is further from the optimum than the scale of the derivative. Additionally, we propose new criteria for pixel selection which are strongly based on in- formation theory, and are faster to compute. We verify these propositions for two popular image difference measures by examining their behavior as the transformation parameters are varied, and by registering a number of typical images.

Better correspondence by registration

Accurate image correspondence is crucial for estimating multiple-view geometry In this paper, we present a registration-based method for improving accuracy of the image correspondences We apply the method to fundamental matrix estimation under practical situations where there are both erroneous matches (outliers) and small feature location errors Our registration-based method can correct feature locational error to less than 0.1 pixel, remedying localization inaccuracy due to feature detectors Moreover, we carefully examine feature similarity based on their post-alignment appearance, providing a more reasonable prior for subsequent outlier detection Experiments show that we can improve feature localization accuracy of the MSER feature detector, which recovers the most accurate feature localization as reported in a recent study by Haja and others As a result of applying our method, we recover the fundamental matrix with better accuracy and more efficiency.