Shape-based averaging.

A new method for averaging multidimensional images is presented, which is based on signed Euclidean distance maps computed for each of the pixel values. We refer to the algorithm as "shape-based averaging" (SBA) because of its similarity to Raya and Udupa's shape-based interpolation method. The new method does not introduce pixel intensities that were not present in the input data, which makes it suitable for averaging nonnumerical data such as label maps (segmentations). Using segmented human brain magnetic resonance images, SBA is compared to label voting for the purpose of averaging image segmentations in a multiclassifier fashion. SBA, on average, performed as well as label voting in terms of recognition rates of the averaged segmentations. SBA produced more regular and contiguous structures with less fragmentation than did label voting. SBA also was more robust for small numbers of atlases and for low atlas resolutions, in particular, when combined with shape-based interpolation. We conclude that SBA improves the contiguity and accuracy of averaged image segmentations.

Virtual 3D planning and guidance of mandibular distraction osteogenesis.

We present a system for 3D planning and pre-operative rehearsal of mandibular distraction osteogenesis procedures. Two primary architectural components are described: a planning system that allows geometric bone manipulation to rapidly explore various modifications and configurations, and a visuohaptic simulator that allows both general-purpose training and preoperative, patient-specific procedure rehearsal. We provide relevant clinical background, then describe the underlying simulation algorithms and their application to craniofacial procedures.

A method for bronchoscope tracking by combining a position sensor and image registration.

This paper describes a method for tracking a bronchoscope by combining a position sensor and image registration. A bronchoscopy guidance system is a tool for providing real-time navigation information acquired from pre-operative CT images to a physician during a bronchoscopic examination. In this system, one of the fundamental functions is tracking a bronchoscope's camera motion. Recently, a very small electromagnetic position sensor has become available. It is possible to insert this sensor into a bronchoscope's working channel to obtain the bronchoscope's camera motion. However, the accuracy of its output is inadequate for bronchoscope tracking. The proposed combination of the sensor and image registration between real and virtual bronchoscopic images derived from CT images is quite useful for improving tracking accuracy. Furthermore, this combination has enabled us to achieve a real-time bronchoscope guidance system. We performed evaluation experiments for the proposed method using a rubber phantom model. The experimental results showed that the proposed system allowed the bronchoscope's camera motion to be tracked at 2.5 frames per second.

Unwarping confocal microscopy images of bee brains by nonrigid registration to a magnetic resonance microscopy image.

Confocal microscopy (CM) is a powerful image acquisition technique that is well established in many biological applications. It provides 3-D acquisition with high spatial resolution and can acquire several different channels of complementary image information. Due to the specimen extraction and preparation process, however, the shapes of imaged objects may differ considerably from their in vivo appearance. Magnetic resonance microscopy (MRM) is an evolving variant of magnetic resonance imaging, which achieves microscopic resolutions using a high magnetic field and strong magnetic gradients. Compared to CM imaging, MRM allows for in situ imaging and is virtually free of geometrical distortions. We propose to combine the advantages of both methods by unwarping CM images using a MRM reference image. Our method incorporates a sequence of image processing operators applied to the MRM image, followed by a two-stage intensity-based registration to compute a nonrigid coordinate transformation between the CM images and the MRM image. We present results obtained using CM images from the brains of 20 honey bees and a MRM image of an in situ bee brain.

Progressive attenuation fields: fast 2D-3D image registration without precomputation.

Computation of digitally reconstructed radiograph (DRR) images is the rate-limiting step in most current intensity-based algorithms for the registration of three-dimensional (3D) images to two-dimensional (2D) projection images. This paper introduces and evaluates the progressive attenuation field (PAF), which is a new method to speed up DRR computation. A PAF is closely related to an attenuation field (AF). A major difference is that a PAF is constructed on the fly as the registration proceeds; it does not require any precomputation time, nor does it make any prior assumptions of the patient pose or limit the permissible range of patient motion. A PAF effectively acts as a cache memory for projection values once they are computed, rather than as a lookup table for precomputed projections like standard AFs. We use a cylindrical attenuation field parametrization, which is better suited for many medical applications of 2D-3D registration than the usual two-plane parametrization. The computed attenuation values are stored in a hash table for time-efficient storage and access. Using clinical gold-standard spine image data sets from five patients, we demonstrate consistent speedups of intensity-based 2D-3D image registration using PAF DRRs by a factor of 10 over conventional ray casting DRRs with no decrease of registration accuracy or robustness.

Fast generation of digitally reconstructed radiographs using attenuation fields with application to 2D-3D image registration.

Generation of digitally reconstructed radiographs (DRRs) is computationally expensive and is typically the rate-limiting step in the execution time of intensity-based two-dimensional to three-dimensional (2D-3D) registration algorithms. We address this computational issue by extending the technique of light field rendering from the computer graphics community. The extension of light fields, which we call attenuation fields (AFs), allows most of the DRR computation to be performed in a preprocessing step; after this precomputation step, DRRs can be generated substantially faster than with conventional ray casting. We derive expressions for the physical sizes of the two planes of an AF necessary to generate DRRs for a given X-ray camera geometry and all possible object motion within a specified range. Because an AF is a ray-based data structure, it is substantially more memory efficient than a huge table of precomputed DRRs because it eliminates the redundancy of replicated rays. Nonetheless, an AF can require substantial memory, which we address by compressing it using vector quantization. We compare DRRs generated using AFs (AF-DRRs) to those generated using ray casting (RC-DRRs) for a typical C-arm geometry and computed tomography images of several anatomic regions. They are quantitatively very similar: the median peak signal-to-noise ratio of AF-DRRs versus RC-DRRs is greater than 43 dB in all cases. We perform intensity-based 2D-3D registration using AF-DRRs and RC-DRRs and evaluate registration accuracy using gold-standard clinical spine image data from four patients. The registration accuracy and robustness of the two methods is virtually identical whereas the execution speed using AF-DRRs is an order of magnitude faster.

Markerless real-time 3-D target region tracking by motion backprojection from projection images.

Accurate and fast localization of a predefined target region inside the patient is an important component of many image-guided therapy procedures. This problem is commonly solved by registration of intraoperative 2-D projection images to 3-D preoperative images. If the patient is not fixed during the intervention, the 2-D image acquisition is repeated several times during the procedure, and the registration problem can be cast instead as a 3-D tracking problem. To solve the 3-D problem, we propose in this paper to apply 2-D region tracking to first recover the components of the transformation that are in-plane to the projections. The 2-D motion estimates of all projections are backprojected into 3-D space, where they are then combined into a consistent estimate of the 3-D motion. We compare this method to intensity-based 2-D to 3-D registration and a combination of 2-D motion backprojection followed by a 2-D to 3-D registration stage. Using clinical data with a fiducial marker-based gold-standard transformation, we show that our method is capable of accurately tracking vertebral targets in 3-D from 2-D motion measured in X-ray projection images. Using a standard tracking algorithm (hyperplane tracking), tracking is achieved at video frame rates but fails relatively often (32% of all frames tracked with target registration error (TRE) better than 1.2 mm, 82% of all frames tracked with TRE better than 2.4 mm). With intensity-based 2-D to 2-D image registration using normalized mutual information (NMI) and pattern intensity (PI), accuracy and robustness are substantially improved. NMI tracked 82% of all frames in our data with TRE better than 1.2 mm and 96% of all frames with TRE better than 2.4 mm. This comes at the cost of a reduced frame rate, 1.7 s average processing time per frame and projection device. Results using PI were slightly more accurate, but required on average 5.4 s time per frame. These results are still substantially faster than 2-D to 3-D registration. We conclude that motion backprojection from 2-D motion tracking is an accurate and efficient method for tracking 3-D target motion, but tracking 2-D motion accurately and robustly remains a challenge.

Intensity-based 2D-3D spine image registration incorporating a single fiducial marker.

RATIONALE AND OBJECTIVES: The two-dimensional (2D)-three dimensional (3D) registration of a computed tomography image to one or more x-ray projection images has a number of image-guided therapy applications. In general, fiducial marker-based methods are fast, accurate, and robust, but marker implantation is not always possible, often is considered too invasive to be clinically acceptable, and entails risk. There also is the unresolved issue of whether it is acceptable to leave markers permanently implanted. Intensity-based registration methods do not require the use of markers and can be automated because such geometric features as points and surfaces do not need to be segmented from the images. However, for spine images, intensity-based methods are susceptible to local optima in the cost function and thus need initial transformations that are close to the correct transformation. MATERIALS AND METHODS: In this report, we propose a hybrid similarity measure for 2D-3D registration that is a weighted combination of an intensity-based similarity measure (mutual information) and a point-based measure using one fiducial marker. We evaluate its registration accuracy and robustness by using gold-standard clinical spine image data from four patients. RESULTS: Mean registration errors for successful registrations for the four patients were 1.3 and 1.1 mm for the intensity-based and hybrid similarity measures, respectively. Whereas the percentage of successful intensity-based registrations (registration error < 2.5 mm) decreased rapidly as the initial transformation got further from the correct transformation, the incorporation of a single marker produced successful registrations more than 99% of the time independent of the initial transformation. CONCLUSION: The use of one fiducial marker reduces 2D-3D spine image registration error slightly and improves robustness substantially. The findings are potentially relevant for image-guided therapy. If one marker is sufficient to obtain clinically acceptable registration accuracy and robustness, as the preliminary results using the proposed hybrid similarity measure suggest, the marker can be placed on a spinous process, which could be accomplished without penetrating muscle or using fluoroscopic guidance, and such a marker could be removed relatively easily.

Modeling liver motion and deformation during the respiratory cycle using intensity-based nonrigid registration of gated MR images.

We present a technique for modeling liver motion during the respiratory cycle using intensity-based nonrigid registration of gated magnetic resonance (MR) images. Three-dimensional MR images of the abdomens of four volunteers were acquired at end-inspiration, end-expiration, and eight time points in between using respiratory gating. The deformation fields between the images were computed using intensity-based rigid and nonrigid registration algorithms. Global motion is modeled by a rigid transformation while local motion is modeled by a free-form deformation based on B-splines. Much of the liver motion was cranial-caudal translation, which was captured by the rigid transformation. However, there was still substantial residual deformation (approximately 10 mm averaged over the entire liver in four volunteers, and 34 mm at one place in the liver of one volunteer). The computed organ motion model can potentially be used to determine an appropriate respiratory-gated radiotherapy window during which the position of the target is known within a specified excursion.

Effect of changing patient position from supine to prone on the accuracy of a Brown-Roberts-Wells stereotactic head frame system.

OBJECTIVE: Despite the growing popularity of frameless image-guided surgery systems, stereotactic frame systems are widely accepted by neurosurgeons and are commonly used to perform biopsies, functional procedures, and stereotactic radiosurgery. We investigated the accuracy of the Brown-Roberts-Wells stereotactic frame system when the mechanical load on the frame changes between preoperative imaging and the intervention because of different patient position: supine during imaging, prone during intervention. METHODS: We analyzed computed tomographic images acquired from 14 patients who underwent stereotactic biopsy, deep brain stimulator implantation, or radiosurgery. Two images were acquired for each patient, one with the patient in the supine position and one in the prone position. The prone images were registered to the respective supine images by use of an intensity-based registration algorithm, once using only the frame and once using only the head. The difference between the transformations produced by these two registrations describes the movement of the patient's head with respect to the frame. RESULTS: The maximum frame-based registration error between the supine and prone positions was 2.8 mm; it was more than 2 mm in two patients and more than 1.5 mm in six patients. Anteroposterior translation is the dominant component of the difference transformation for most patients. In general, the magnitude of the movement increased with brain volume, which is an index of head weight. CONCLUSION: To minimize frame-based registration error caused by a change in the mechanical load on the frame, stereotactic procedures should be performed with the patient in the identical position during imaging and intervention.

Designing optically tracked instruments for image-guided surgery.

Most image-guided surgery (IGS) systems track the positions of surgical instruments in the physical space occupied by the patient. This task is commonly performed using an optical tracking system that determines the positions of fiducial markers such as infrared-emitting diodes or retroreflective spheres that are attached to the instrument. Instrument tracking error is an important component of the overall IGS system error. This paper is concerned with the effect of fiducial marker configuration (number and spatial distribution) on tip position tracking error. Statistically expected tip position tracking error is calculated by applying results from the point-based registration error theory developed by Fitzpatrick et al. Tracking error depends not only on the error in localizing the fiducials, which is the error value generally provided by manufacturers of optical tracking systems, but also on the number and spatial distribution of the tracking fiducials and the position of the instrument tip relative to the fiducials. The theory is extended in two ways. First, a formula is derived for the special case in which the fiducials and the tip are collinear. Second, the theory is extended for the case in which there is a composition of transformations, as is the situation for tracking an instrument relative to a coordinate reference frame (i.e., a set of fiducials attached to the patient). The derivation reveals that the previous theory may be applied independently to the two transformations; the resulting independent components of tracking error add in quadrature to give the overall tracking error. The theoretical results are verified with numerical simulations and experimental measurements. The results in this paper may be useful for the design of optically tracked instruments for image-guided surgery; this is illustrated with several examples.

Performance-based classifier combination in atlas-based image segmentation using expectation-maximization parameter estimation.

It is well known in the pattern recognition community that the accuracy of classifications obtained by combining decisions made by independent classifiers can be substantially higher than the accuracy of the individual classifiers. We have previously shown this to be true for atlas-based segmentation of biomedical images. The conventional method for combining individual classifiers weights each classifier equally (vote or sum rule fusion). In this paper, we propose two methods that estimate the performances of the individual classifiers and combine the individual classifiers by weighting them according to their estimated performance. The two methods are multiclass extensions of an expectation-maximization (EM) algorithm for ground truth estimation of binary classification based on decisions of multiple experts (Warfield et al., 2004). The first method performs parameter estimation independently for each class with a subsequent integration step. The second method considers all classes simultaneously. We demonstrate the efficacy of these performance-based fusion methods by applying them to atlas-based segmentations of three-dimensional confocal microscopy images of bee brains. In atlas-based image segmentation, multiple classifiers arise naturally by applying different registration methods to the same atlas, or the same registration method to different atlases, or both. We perform a validation study designed to quantify the success of classifier combination methods in atlas-based segmentation. By applying random deformations, a given ground truth atlas is transformed into multiple segmentations that could result from imperfect registrations of an image to multiple atlas images. In a second evaluation study, multiple actual atlas-based segmentations are combined and their accuracies computed by comparing them to a manual segmentation. We demonstrate in both evaluation studies that segmentations produced by combining multiple individual registration-based segmentations are more accurate for the two classifier fusion methods we propose, which weight the individual classifiers according to their EM-based performance estimates, than for simple sum rule fusion, which weights each classifier equally.

Volume-preserving nonrigid registration of MR breast images using free-form deformation with an incompressibility constraint.

In this paper, we extend a previously reported intensity-based nonrigid registration algorithm by using a novel regularization term to constrain the deformation. Global motion is modeled by a rigid transformation while local motion is described by a free-form deformation based on B-splines. An information theoretic measure, normalized mutual information, is used as an intensity-based image similarity measure. Registration is performed by searching for the deformation that minimizes a cost function consisting of a weighted combination of the image similarity measure and a regularization term. The novel regularization term is a local volume-preservation (incompressibility) constraint, which is motivated by the assumption that soft tissue is incompressible for small deformations and short time periods. The incompressibility constraint is implemented by penalizing deviations of the Jacobian determinant of the deformation from unity. We apply the nonrigid registration algorithm with and without the incompressibility constraint to precontrast and post-contrast magnetic resonance (MR) breast images from 17 patients. Without using a constraint, the volume of contrast-enhancing lesions decreases by 1%-78% (mean 26%). Image improvement (motion artifact reduction) obtained using the new constraint is compared with that obtained using a smoothness constraint based on the bending energy of the coordinate grid by blinded visual assessment of maximum intensity projections of subtraction images. For both constraints, volume preservation improves, and motion artifact correction worsens, as the weight of the constraint penalty term increases. For a given volume change of the contrast-enhancing lesions (2% of the original volume), the incompressibility constraint reduces motion artifacts better than or equal to the smoothness constraint in 13 out of 17 cases (better in 9, equal in 4, worse in 4). The preliminary results suggest that incorporation of the incompressibility regularization term improves intensity-based free-form nonrigid registration of contrast-enhanced MR breast images by greatly reducing the problem of shrinkage of contrast-enhancing structures while simultaneously allowing motion artifacts to be substantially reduced.

Implementation, calibration and accuracy testing of an image-enhanced endoscopy system.

This paper presents a new method for image-guided surgery called image-enhanced endoscopy. Registered real and virtual endoscopic images (perspective volume renderings generated from the same view as the endoscope camera using a preoperative image) are displayed simultaneously; when combined with the ability to vary tissue transparency in the virtual images, this provides surgeons with the ability to see beyond visible surfaces and, thus, provides additional exposure during surgery. A mount with four photoreflective spheres is rigidly attached to the endoscope and its position and orientation is tracked using an optical position sensor. Generation of virtual images that are accurately registered to the real endoscopic images requires calibration of the tracked endoscope. The calibration process determines intrinsic parameters (that represent the projection of three-dimensional points onto the two-dimensional endoscope camera imaging plane) and extrinsic parameters (that represent the transformation from the coordinate system of the tracker mount attached to the endoscope to the coordinate system of the endoscope camera), and determines radial lens distortion. The calibration routine is fast, automatic, accurate and reliable, and is insensitive to rotational orientation of the endoscope. The routine automatically detects, localizes, and identifies dots in a video image snapshot of the calibration target grid and determines the calibration parameters from the sets of known physical coordinates and localized image coordinates of the target grid dots. Using nonlinear lens-distortion correction, which can be performed at real-time rates (30 frames per second), the mean projection error is less than 0.5 mm at distances up to 25 mm from the endoscope tip, and less than 1.0 mm up to 45 mm. Experimental measurements and point-based registration error theory show that the tracking error is about 0.5-0.7 mm at the tip of the endoscope and less than 0.9 mm for all points in the field of view of the endoscope camera at a distance of up to 65 mm from the tip. It is probable that much of the projection error is due to endoscope tracking error rather than calibration error. Two examples of clinical applications are presented to illustrate the usefulness of image-enhanced endoscopy. This method is a useful addition to conventional image-guidance systems, which generally show only the position of the tip (and sometimes the orientation) of a surgical instrument or probe on reformatted image slices.

The impact of fiducial distribution on headset-based registration in image-guided sinus surgery.

OBJECTIVE: The objective of this study was to assess registration error due to fiducial configuration for the ENT headsets for the CBYON Suite (CBYON, Mountain View, CA) and InstaTrak (GEMS Navigation and Visualization, Waukesha, WI). STUDY DESIGN: Axial CT scans (1-mm slice thickness) were obtained of for 24 cadaveric heads using the CBYON headset and for 23 cadaveric heads using the GEMS headset. The CBYON and GEMS NAV software were used to calculate the fiducial registration error (FRE). Fiducial localization error (FLE) was estimated from FRE. Theoretical target registration error (TRE) was calculated at 11 targets. RESULTS: The FRE for CBYON and GEMS NAV was 0.69 mm and 0.27 mm, respectively. The theoretical TRE for CBYON and GEMS NAV was 0.41 mm and 0.30 mm, respectively. The theoretical TRE was greater at targets posterior in the sinus cavities. CONCLUSION: Theoretical TRE values for both ENT headsets are less than clinically observed TRE. Clinically observed TRE is likely due to repositioning accuracy. EBM RATING: B-2.

Nonrigid image registration in shared-memory multiprocessor environments with application to brains, breasts, and bees.

One major problem with nonrigid 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., intraoperative imaging. This paper presents a parallel implementation of a nonrigid image registration algorithm. It takes advantage of shared-memory multiprocessor computer architectures using multithreaded programming by partitioning of data and partitioning of tasks, depending on the computational subproblem. For three different biomedical applications (intraoperative brain deformation, contrast-enhanced MR mammography, intersubject brain registration), the scaling behavior of the algorithm is quantitatively analyzed. The method is demonstrated to perform the computation of intra-operative brain deformation in less than a minute using 64 CPUs on a 128-CPU shared-memory supercomputer (SGI Origin 3800). It is shown that its serial component is no more than 2% 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). The parallel implementation of our algorithm is, therefore, capable of solving nonrigid 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.

Selective image similarity measure for bronchoscope tracking based on image registration.

We propose a selective method of measurement for computing image similarities based on characteristic structure extraction and demonstrate its application to flexible endoscope navigation, in particular to a bronchoscope navigation system. Camera motion tracking is a fundamental function required for image-guided treatment or therapy systems. In recent years, an ultra-tiny electromagnetic sensor commercially became available, and many image-guided treatment or therapy systems use this sensor for tracking the camera position and orientation. However, due to space limitations, it is difficult to equip the tip of a bronchoscope with such a position sensor, especially in the case of ultra-thin bronchoscopes. Therefore, continuous image registration between real and virtual bronchoscopic images becomes an efficient tool for tracking the bronchoscope. Usually, image registration is done by calculating the image similarity between real and virtual bronchoscopic images. Since global schemes to measure image similarity, such as mutual information, squared gray-level difference, or cross correlation, average differences in intensity values over an entire region, they fail at tracking of scenes where less characteristic structures can be observed. The proposed method divides an entire image into a set of small subblocks and only selects those in which characteristic shapes are observed. Then image similarity is calculated within the selected subblocks. Selection is done by calculating feature values within each subblock. We applied our proposed method to eight pairs of chest X-ray CT images and bronchoscopic video images. The experimental results revealed that bronchoscope tracking using the proposed method could track up to 1600 consecutive bronchoscopic images (about 50s) without external position sensors. Tracking performance was greatly improved in comparison with a standard method utilizing squared gray-level differences of the entire images.

A study of the accuracy of cyberknife spinal radiosurgery using skeletal structure tracking.

OBJECTIVE: New technology has enabled the increasing use of radiosurgery to ablate spinal lesions. The first generation of the CyberKnife (Accuray, Inc., Sunnyvale, CA) image-guided radiosurgery system required implanted radiopaque markers (fiducials) to localize spinal targets. A recently developed and now commercially available spine tracking technology called Xsight (Accuray, Inc.) tracks skeletal structures and eliminates the need for implanted fiducials. The Xsight system localizes spinal targets by direct reference to the adjacent vertebral elements. This study sought to measure the accuracy of Xsight spine tracking and provide a qualitative assessment of overall system performance. METHODS: Total system error, which is defined as the distance between the centroids of the planned and delivered dose distributions and represents all possible treatment planning and delivery errors, was measured using a realistic, anthropomorphic head-and-neck phantom. The Xsight tracking system error component of total system error was also computed by retrospectively analyzing image data obtained from eleven patients with a total of 44 implanted fiducials who underwent CyberKnife spinal radiosurgery. RESULTS: The total system error of the Xsight targeting technology was measured to be 0.61 mm. The tracking system error component was found to be 0.49 mm. CONCLUSION: The Xsight spine tracking system is practically important because it is accurate and eliminates the use of implanted fiducials. Experience has shown this technology to be robust under a wide range of clinical circumstances.

Comprehensive maps of Drosophila higher olfactory centers: spatially segregated fruit and pheromone representation.

In Drosophila, approximately 50 classes of olfactory receptor neurons (ORNs) send axons to 50 corresponding glomeruli in the antennal lobe. Uniglomerular projection neurons (PNs) relay olfactory information to the mushroom body (MB) and lateral horn (LH). Here, we combine single-cell labeling and image registration to create high-resolution, quantitative maps of the MB and LH for 35 input PN channels and several groups of LH neurons. We find (1) PN inputs to the MB are stereotyped as previously shown for the LH; (2) PN partners of ORNs from different sensillar groups are clustered in the LH; (3) fruit odors are represented mostly in the posterior-dorsal LH, whereas candidate pheromone-responsive PNs project to the anterior-ventral LH; (4) dendrites of single LH neurons each overlap with specific subsets of PN axons. Our results suggest that the LH is organized according to biological values of olfactory input.