Improve Registration Accuracy Research Articles

MIND Demons: Symmetric Diffeomorphic Deformable Registration of MR and CT for Image-Guided Spine Surgery.

Intraoperative localization of target anatomy and critical structures defined in preoperative MR/CT images can be achieved through the use of multimodality deformable registration. We propose a symmetric diffeomorphic deformable registration algorithm incorporating a modality-independent neighborhood descriptor (MIND) and a robust Huber metric for MR-to-CT registration. The method, called MIND Demons, finds a deformation field between two images by optimizing an energy functional that incorporates both the forward and inverse deformations, smoothness on the integrated velocity fields, a modality-insensitive similarity function suitable to multimodality images, and smoothness on the diffeomorphisms themselves. Direct optimization without relying on the exponential map and stationary velocity field approximation used in conventional diffeomorphic Demons is carried out using a Gauss-Newton method for fast convergence. Registration performance and sensitivity to registration parameters were analyzed in simulation, phantom experiments, and clinical studies emulating application in image-guided spine surgery, and results were compared to mutual information (MI) free-form deformation (FFD), local MI (LMI) FFD, normalized MI (NMI) Demons, and MIND with a diffusion-based registration method (MIND-elastic). The method yielded sub-voxel invertibility (0.008 mm) and nonzero-positive Jacobian determinants. It also showed improved registration accuracy in comparison to the reference methods, with mean target registration error (TRE) of 1.7 mm compared to 11.3, 3.1, 5.6, and 2.4 mm for MI FFD, LMI FFD, NMI Demons, and MIND-elastic methods, respectively. Validation in clinical studies demonstrated realistic deformations with sub-voxel TRE in cases of cervical, thoracic, and lumbar spine.

Effects of Resolution and Registration Algorithm on the Accuracy of EPI vNavs for Real Time Head Motion Correction in MRI.

Low-resolution, EPI-based Volumetric Navigators (vNavs) have been used as a prospective motion-correction system in a variety of MRI neuroimaging pulse sequences. The use of low-resolution volumes represents a trade-off between motion tracking accuracy and acquisition time. However, this means that registration must be accurate on the order of 0.2 voxels or less to be effective for motion correction. While vNavs have shown promising results in clinical and research use, the choice of navigator and registration algorithm have not previously been systematically evaluated. In this work we experimentally evaluate the accuracy of vNavs, and possible design choices for future improvements to the system, using real human data. We acquired navigator volumes at three isotropic resolutions (6.4 mm, 8 mm, and 10 mm) with known rotations and translations. The vNavs were then rigidly registered using trilinear, tricubic, and cubic B-spline interpolation. We demonstrate a novel refactoring of the cubic B-spline algorithm that stores pre-computed coefficients to reduce the per-interpolation time to be identical to tricubic interpolation. Our results show that increasing vNav resolution improves registration accuracy, and that cubic B-splines provide the highest registration accuracy at all vNav resolutions. Our results also suggest that the time required by vNavs may be reduced by imaging at 10 mm resolution, without substantial cost in registration accuracy.

Integrating Structural MRI and DTI in Brain Deformation Simulation for Image-Guided Neurosurgery

In image-guided neurosurgery, pre-operative magnetic resonance (MR) images can be fused with intra-operative ones to enhance visualization and navigation to determine the resection margin. However, intra-operative brain deformation, after opening the skull, greatly degrades accurate matching between these images. Although direct registering these images aims to address such a deformation problem, simulating brain shifts can potentially improve registration accuracy. In this paper, we introduce the biomechanical finite element model (FEM) to compute brain deformation such as parenchyma, tumor, ventricle and white matter shifts during craniotomy using both MRI and diffusion tensor images (DTI). In experiments, we show the pipeline for brain shift simulation and compare the results using different loading and boundary conditions of the FEM model. This method can be potentially used for simulating brain shifts before non-rigid image registration between intra-operative and pre-operative images in image-guided neurosurgery. Keywords: Brain deformation, diffusion tenser imaging (DTI), finite-element model, image guided neurosurgery, magnetic resonance (MR), neurosurgery.

Evaluation of model-based deformation correction in image-guided liver surgery via tracked intraoperative ultrasound.

Soft-tissue deformation represents a significant error source in current surgical navigation systems used for open hepatic procedures. While numerous algorithms have been proposed to rectify the tissue deformation that is encountered during open liver surgery, clinical validation of the proposed methods has been limited to surface-based metrics, and subsurface validation has largely been performed via phantom experiments. The proposed method involves the analysis of two deformation-correction algorithms for open hepatic image-guided surgery systems via subsurface targets digitized with tracked intraoperative ultrasound (iUS). Intraoperative surface digitizations were acquired via a laser range scanner and an optically tracked stylus for the purposes of computing the physical-to-image space registration and for use in retrospective deformation-correction algorithms. Upon completion of surface digitization, the organ was interrogated with a tracked iUS transducer where the iUS images and corresponding tracked locations were recorded. Mean closest-point distances between the feature contours delineated in the iUS images and corresponding three-dimensional anatomical model generated from preoperative tomograms were computed to quantify the extent to which the deformation-correction algorithms improved registration accuracy. The results for six patients, including eight anatomical targets, indicate that deformation correction can facilitate reduction in target error of [Formula: see text].

MIND Demons for MR-to-CT Deformable Image Registration In Image-Guided Spine Surgery.

Localization of target anatomy and critical structures defined in preoperative MR images can be achieved by means of multi-modality deformable registration to intraoperative CT. We propose a symmetric diffeomorphic deformable registration algorithm incorporating a modality independent neighborhood descriptor (MIND) and a robust Huber metric for MR-to-CT registration. The method, called MIND Demons, solves for the deformation field between two images by optimizing an energy functional that incorporates both the forward and inverse deformations, smoothness on the velocity fields and the diffeomorphisms, a modality-insensitive similarity function suitable to multi-modality images, and constraints on geodesics in Lagrangian coordinates. Direct optimization (without relying on an exponential map of stationary velocity fields used in conventional diffeomorphic Demons) is carried out using a Gauss-Newton method for fast convergence. Registration performance and sensitivity to registration parameters were analyzed in simulation, in phantom experiments, and clinical studies emulating application in image-guided spine surgery, and results were compared to conventional mutual information (MI) free-form deformation (FFD), local MI (LMI) FFD, and normalized MI (NMI) Demons. The method yielded sub-voxel invertibility (0.006 mm) and nonsingular spatial Jacobians with capability to preserve local orientation and topology. It demonstrated improved registration accuracy in comparison to the reference methods, with mean target registration error (TRE) of 1.5 mm compared to 10.9, 2.3, and 4.6 mm for MI FFD, LMI FFD, and NMI Demons methods, respectively. Validation in clinical studies demonstrated realistic deformation with sub-voxel TRE in cases of cervical, thoracic, and lumbar spine. A modality-independent deformable registration method has been developed to estimate a viscoelastic diffeomorphic map between preoperative MR and intraoperative CT. The method yields registration accuracy suitable to application in image-guided spine surgery across a broad range of anatomical sites and modes of deformation.

Joint registration of ultrasound, CT and a shape+pose statistical model of the lumbar spine for guiding anesthesia.

Facet joint injections and epidural needle insertions are widely used for spine anesthesia. Accurate needle placement is important for effective therapy delivery and avoiding complications arising from damage of soft tissue and nerves. Needle guidance is usually performed by fluoroscopy or palpation, resulting in radiation exposure and multiple needle re-insertions. Several ultrasound (US)-based approaches have been proposed but have not found wide acceptance in clinical routine. This is mainly due to difficulties in interpretation of the complex spinal anatomy in US, which leads to clinicians' lack of confidence in relying only on information derived from US for needle guidance. We introduce a multimodal joint registration technique that takes advantage of easy-to-interpret preprocedure computed topography (CT) scans of the lumbar spine to concurrently register a shape+pose model to the intraprocedure 3D US. Common shape coefficients are assumed between two modalities, while pose coefficients are specific to each modality. The joint method was evaluated on patient data consisting of ten pairs of US and CT scans of the lumbar spine. It was successfully applied in all cases and yielded an RMS shape error of 2.1mm compared to the CT ground truth. The joint registration technique was compared to a previously proposed method of statistical model to US registrationRasoulian etal. (Information processing in computer-assisted interventions. Springer, Berlin, pp 51-60, 2013). The joint framework improved registration accuracy to US in 7 out of 17 visible vertebrae, belonging to four patients. In the remaining cases, the two methods were equally accurate. The joint registration allows visualization and augmentation of important anatomy in both the US and CT domain and improves the registration accuracy in both modalities. Observing the patient-specific model in the CT domain allows the clinicians to assess the local registration accuracy qualitatively, which is likely to increase their confidence in using the US model for deriving needle guidance decisions.

Simultaneous MV-kV imaging for intrafractional motion management during volumetric-modulated arc therapy delivery.

The purpose of this study was to evaluate the accuracy and clinical feasibility of a motion monitoring method employing simultaneously acquired MV and kV images during volumetric‐modulated arc therapy (VMAT). Short‐arc digital tomosynthesis (SA‐DTS) is used to improve the quality of the MV images that are then combined with orthogonally acquired kV images to assess 3D motion. An anthropomorphic phantom with implanted gold seeds was used to assess accuracy of the method under static, typical prostatic, and respiratory motion scenarios. Automatic registration of kV images and single MV frames or MV SA‐DTS reconstructed with arc lengths from 2° to 7° with the appropriate reference fiducial template images was performed using special purpose‐built software. Clinical feasibility was evaluated by retrospectively analyzing images acquired over four or five sessions for each of three patients undergoing hypofractionated prostate radiotherapy. The standard deviation of the registration error in phantom using MV SA‐DTS was similar to single MV images for the static and prostate motion scenarios (σ=0.25 mm). Under respiratory motion conditions, the standard deviation of the registration error increased to 0.7 mm and 1.7 mm for single MV and MV SA‐DTS, respectively. Registration failures were observed with the respiratory scenario only and were due to motion‐induced fiducial blurring. For the three patients studied, the mean and standard deviation of the difference between automatic registration using 4° MV SA‐DTS and manual registration using single MV images results was 0.07±0.52 mm. The MV SA‐DTS results in patients were, on average, superior to single‐frame MV by nearly 1 mm — significantly more than what was observed in phantom. The best MV SA‐DTS results were observed with arc lengths of 3° to 4°. Registration failures in patients using MV SA‐DTS were primarily due to blockage of the gold seeds by the MLC. The failure rate varied from 2% to 16%. Combined MV SA‐DTS and kV imaging is feasible for intratreatment motion monitoring during VMAT of anatomic sites where limited motion is expected, and improves registration accuracy compared to single MV/kV frames. To create a clinically robust technique, further improvements to ensure visualization of fiducials at the desired control points without degradation of the treatment plan are needed.PACS number(s): 87.55.km, 87.55.N‐

Gradient Enhanced Image Pyramid for Improved Nonlinear Image Registration

In this paper we investigate the use of image pyramid within a hierarchical registration framework for improved nonlinear image registration. Gaussian image pyramid is commonly used to reduce image complexity so that registration can be performed in a coarse to fine manner. In this paper, we apply two edge preserving filters, the bilateral filter and the guided filter, to generate image pyramids that can preserve strong gradient so as to improve registration accuracy. In addition, we propose a bilateral fractional differential based image enhancement filter and combine its output with a guided filter to generate another image pyramid that further enhances the gradient of strong image components. Registration is performed within a hierarchical framework where the model complexity of a Discrete Cosine Transformation (DCT) based nonlinear model is increased to couple with the image pyramid. Different image pyramids are compared by using three types of synthetic deformation fields. Experimental results show that registration using the gradient enhanced image pyramid achieves more accurate registration than registration using the Gaussian image pyramid.

Progressive internal landmark registration for surgical navigation in laparoscopic gastrectomy for gastric cancer.

A surgical navigation system supports the comprehension of anatomical information during surgery. Patient-to-image registration is the alignment process between CT volume and patient coordinate systems. Achieving accurate registration in the surgical navigation of laparoscopic surgery is very challenging due to soft tissue deformation. This paper presents a new patient-to-image registration method based on internal anatomical landmarks for improving registration accuracy in the surgical navigation of laparoscopic gastrectomy for gastric cancer. Our proposed registration method progressively utilizes internal anatomical landmarks. In laparoscopic gastrectomy for gastric cancer, the surgeon cuts the blood vessels around the stomach. The positions of the cut vessels are sequentially used as fiducials for registration during surgery. The proposed method uses a weighted point-based registration method for computing the transformation matrix using the fiducials both on the body surface and on the blood vessels. When a blood vessel is cut during surgery, the proposed progressive registration method measures the cut vessel's position and computes a transformation matrix by adding the cut vessel as a fiducial. We applied our proposed progressive registration method using the positional information of the blood vessels acquired during laparoscopic gastrectomy in 20 cases. We evaluated it using target registration error in four blood vessels. The average target registration error in the four blood vessels was 12.6mm and ranged from 2.1 to 32.9mm. Since the proposed progressive registration can reduce registration error, our proposed method is very useful for the surgical navigation of laparoscopic gastrectomy. Our proposed progressive registration method might increase the accuracy of surgical navigation in laparoscopic gastrectomy.

Development of a registration framework to validate MRI with histology for prostate focal therapy.

Focal therapy has been proposed as an alternative method to whole-gland treatment for prostate cancer when aiming to reduce treatment side effects. The authors recently validated a radiobiological model which takes into account tumor location and tumor characteristics including tumor cell density, Gleason score, and hypoxia in order to plan optimal dose distributions for focal therapy. The authors propose that this model can be informed using multiparametric MRI (mpMRI) and in this study present a registration framework developed to map prostate mpMRI and histology data, where histology will provide the "ground truth" data regarding tumor location and biology. The authors aim to apply this framework to a growing database to develop a prostate biological atlas which will enable MRI based planning for prostate focal therapy treatment. Six patients scheduled for routine radical prostatectomy were used in this proof-of-concept study. Each patient underwent mpMRI scanning prior to surgery, after which the excised prostate specimen was formalin fixed and mounted in agarose gel in a custom designed sectioning box. T2-weighted MRI of the specimen in the sectioning box was acquired, after which 5 mm sections of the prostate were cut and histology sections were microtomed. A number of image processing and registration steps were used to register histology images with ex vivo MRI and deformable image registration (DIR) was applied to 3D T2w images to align the in vivo and ex vivo MRI data. Dice coefficient metrics and corresponding feature points from two independent annotators were selected in order to assess the DIR accuracy. Images from all six patients were registered, providing histology and in vivo MRI in the ex vivo MRI frame of reference for each patient. Results demonstrated that their DIR methodology to register in vivo and ex vivo 3D T2w MRI improved accuracy in comparison with an initial manual alignment for prostates containing features which were readily visible on MRI. The average estimated uncertainty between in vivo MRI and histology was 3.3 mm, which included an average error of 3.1 mm between in vivo and ex vivo MRI after applying DIR. The mean dice coefficient for the prostate contour between in vivo and ex vivo MRI increased from 0.83 before DIR to 0.93 after DIR. The authors have developed a registration framework for mapping in vivo MRI data of the prostate with histology by implementing a number of processing steps and ex vivo MRI of the prostate specimen. Validation of DIR was challenging, particularly in prostates with few or mostly linear rather than spherical shaped features. Refinement of their MR imaging protocols to improve the data quality is currently underway which may improve registration accuracy. Additional mpMRI sequences will be registered within this framework to quantify prostate tumor location and biology.

3D nonrigid medical image registration using a new information theoretic measure

This work presents a novel method for the nonrigid registration of medical images based on the Arimoto entropy, a generalization of the Shannon entropy. The proposed method employed the Jensen–Arimoto divergence measure as a similarity metric to measure the statistical dependence between medical images. Free-form deformations were adopted as the transformation model and the Parzen window estimation was applied to compute the probability distributions. A penalty term is incorporated into the objective function to smooth the nonrigid transformation. The goal of registration is to optimize an objective function consisting of a dissimilarity term and a penalty term, which would be minimal when two deformed images are perfectly aligned using the limited memory BFGS optimization method, and thus to get the optimal geometric transformation. To validate the performance of the proposed method, experiments on both simulated 3D brain MR images and real 3D thoracic CT data sets were designed and performed on the open source elastix package. For the simulated experiments, the registration errors of 3D brain MR images with various magnitudes of known deformations and different levels of noise were measured. For the real data tests, four data sets of 4D thoracic CT from four patients were selected to assess the registration performance of the method, including ten 3D CT images for each 4D CT data covering an entire respiration cycle. These results were compared with the normalized cross correlation and the mutual information methods and show a slight but true improvement in registration accuracy.

Deformable image registration by combining uncertainty estimates from supervoxel belief propagation.

Discrete optimisation strategies have a number of advantages over their continuous counterparts for deformable registration of medical images. For example: it is not necessary to compute derivatives of the similarity term; dense sampling of the search space reduces the risk of becoming trapped in local optima; and (in principle) an optimum can be found without resorting to iterative coarse-to-fine warping strategies. However, the large complexity of high-dimensional medical data renders a direct voxel-wise estimation of deformation vectors impractical. For this reason, previous work on medical image registration using graphical models has largely relied on using a parameterised deformation model and on the use of iterative coarse-to-fine optimisation schemes. In this paper, we propose an approach that enables accurate voxel-wise deformable registration of high-resolution 3D images without the need for intermediate image warping or a multi-resolution scheme. This is achieved by representing the image domain as multiple comprehensive supervoxel layers and making use of the full marginal distribution of all probable displacement vectors after inferring regularity of the deformations using belief propagation. The optimisation acts on the coarse scale representation of supervoxels, which provides sufficient spatial context and is robust to noise in low contrast areas. Minimum spanning trees, which connect neighbouring supervoxels, are employed to model pair-wise deformation dependencies. The optimal displacement for each voxel is calculated by considering the probabilities for all displacements over all overlapping supervoxel graphs and subsequently seeking the mode of this distribution. We demonstrate the applicability of this concept for two challenging applications: first, for intra-patient motion estimation in lung CT scans; and second, for atlas-based segmentation propagation of MRI brain scans. For lung registration, the voxel-wise mode of displacements is found using the mean-shift algorithm, which enables us to determine continuous valued sub-voxel motion vectors. Finding the mode of brain segmentation labels is performed using a voxel-wise majority voting weighted by the displacement uncertainty estimates. Our experimental results show significant improvements in registration accuracy when using the additional information provided by the registration uncertainty estimates. The multi-layer approach enables fusion of multiple complementary proposals, extending the popular fusion approaches from multi-image registration to probabilistic one-to-one image registration.

Robust anatomical landmark detection with application to MR brain image registration

Comparison of human brain MR images is often challenged by large inter-subject structural variability. To determine correspondences between MR brain images, most existing methods typically perform a local neighborhood search, based on certain morphological features. They are limited in two aspects: (1) pre-defined morphological features often have limited power in characterizing brain structures, thus leading to inaccurate correspondence detection, and (2) correspondence matching is often restricted within local small neighborhoods and fails to cater to images with large anatomical difference. To address these limitations, we propose a novel method to detect distinctive landmarks for effective correspondence matching. Specifically, we first annotate a group of landmarks in a large set of training MR brain images. Then, we use regression forest to simultaneously learn (1) the optimal sets of features to best characterize each landmark and (2) the non-linear mappings from the local patch appearances of image points to their 3D displacements towards each landmark. The learned regression forests are used as landmark detectors to predict the locations of these landmarks in new images. Because each detector is learned based on features that best distinguish the landmark from other points and also landmark detection is performed in the entire image domain, our method can address the limitations in conventional methods. The deformation field estimated based on the alignment of these detected landmarks can then be used as initialization for image registration. Experimental results show that our method is capable of providing good initialization even for the images with large deformation difference, thus improving registration accuracy.

Division-based large point set registration using coherent point drift (CPD) with automatic parameter tuning1

Large point sets consists of unordered sets of usually 3D coordinates representing a surface (e.g., face) or a volume. With the advent of laser scanners the surface can be captured with high resolution generating a large amount of data. Processing this amount of data for point set registration efficiently, poses the type of challenges being addressed by the big data community. Coherent Point Drift (CPD) is a state-of-the-art point set registration method, that is able to handle large point cloud registration in O(n) time with the incorporation of the Fast Gauss Transform (FGT) and low-rank matrix approximation (LRA). However, its registration accuracy degrades rapidly for large point sets. To overcome this, we present a strategy that divides a large point set into several smaller overlapping subsets. These subsets are then independently registered using CPD that are then merged for final registration. To improve registration accuracy, we also propose a method to tune the width parameter of the Gaussian kernel in CPD. The proposed method has been tested on four large datasets, including the USF 3D face dataset. The results show that the proposed method is able to register large datasets with greater speed and accuracy than the state-of-the-art CPD method.

WE-AB-BRA-01: 3D-2D Image Registration for Target Localization in Spine Surgery: Comparison of Similarity Metrics Against Robustness to Content Mismatch

Purpose: In image-guided spine surgery, mapping 3D preoperative images to 2D intraoperative images via 3D-2D registration can provide valuable assistance in target localization. However, the presence of surgical instrumentation, hardware implants, and soft-tissue resection/displacement causes mismatches in image content, confounding existing registration methods. Manual/semi-automatic methods to mask such extraneous content is time consuming, user-dependent, error prone, and disruptive to clinical workflow. We developed and evaluated 2 novel similarity metrics within a robust registration framework to overcome such challenges in target localization. Methods: An IRB-approved retrospective study in 19 spine surgery patients included 19 preoperative 3D CT images and 50 intraoperative mobile radiographs in cervical, thoracic, and lumbar spine regions. A neuroradiologist provided truth definition of vertebral positions in CT and radiography. 3D-2D registration was performed using the CMA-ES optimizer with 4 gradient-based image similarity metrics: (1) gradient information (GI); (2) gradient correlation (GC); (3) a novel variant referred to as gradient orientation (GO); and (4) a second variant referred to as truncated gradient correlation (TGC). Registration accuracy was evaluated in terms of the projection distance error (PDE) of the vertebral levels. Results: Conventional similarity metrics were susceptible to gross registration error and failure modes associated with the presence of surgical instrumentation: for GI, the median PDE and interquartile range was 33.0±43.6 mm; similarly for GC, PDE = 23.0±92.6 mm respectively. The robust metrics GO and TGC, on the other hand, demonstrated major improvement in PDE (7.6 ±9.4 mm and 8.1± 18.1 mm, respectively) and elimination of gross failure modes. Conclusion: The proposed GO and TGC similarity measures improve registration accuracy and robustness to gross failure in the presence of strong image content mismatch. Such registration capability could offer valuable assistance in target localization without disruption of clinical workflow. G. Kleinszig and S. Vogt are employees of Siemens Healthcare.

SU-E-J-101: Improved CT to CBCT Deformable Registration Accuracy by Incorporating Multiple CBCTs

Purpose: Combining prior day CBCT contours with STAPLE was previously shown to improve automated prostate contouring. These accurate STAPLE contours are now used to guide the planning CT to pre-treatment CBCT deformable registration. Methods: Six IGRT prostate patients with daily kilovoltage CBCT had their original planning CT and 9 CBCTs contoured by the same physician. These physician contours for the planning CT and each prior CBCT are deformed to match the current CBCT anatomy, producing multiple contour sets. These sets are then combined using STAPLE into one optimal set (e.g. for day 3 CBCT, combine contours produced using the plan plus day 1 and 2 CBCTs). STAPLE computes a probabilistic estimate of the true contour from this collection of contours by maximizing sensitivity and specificity. The deformation field from planning CT to CBCT registration is then refined by matching its deformed contours to the STAPLE contours. ADMIRE (Elekta Inc.) was used for this. The refinement does not force perfect agreement of the contours, typically Dice's Coefficient (DC) of > 0.9 is obtained, and the image difference metric remains in the optimization of the deformable registration. Results: The average DC between physician delineated CBCT contours and deformed planning CT contours for the bladder, rectum and prostate was 0.80, 0.79 and 0.75, respectively. The accuracy significantly improved to 0.89, 0.84 and 0.84 (P<0.001 for all) when using the refined deformation field. The average time to run STAPLE with five scans and refine the planning CT deformation was 66 seconds on a Telsa K20c GPU. Conclusion: Accurate contours generated from multiple CBCTs provided guidance for CT to CBCT deformable registration, significantly improving registration accuracy as measured by contour DC. A more accurate deformation field is now available for transferring dose or electron density to the CBCT for adaptive planning. Research grant from Elekta

Multimodal image registration with joint structure tensor and local entropy.

Nonrigid registration of multimodal medical images remains a challenge in image-guided interventions. A common approach is to use mutual information (MI), which is robust to the intensity variations across modalities. However, primarily based on intensity distribution, MI does not take into account of underlying spatial and structural information of the images, which might lead to local optimization. To address such a challenge, this paper proposes a two-stage multimodal nonrigid registration scheme with joint structural information and local entropy. In our two-stage multimodal nonrigid registration scheme, both the reference image and floating image are firstly converted to a common space. A unified representation in the common space for the images is constructed by fusing the structure tensor (ST) trace with the local entropy (LE). Through the representation that reflects its geometry uniformly across modalities, the complicated deformation field is estimated using L(1) or L(2) distance. We compared our approach to four other methods: (1) the method using LE, (2) the method using ST, (3) the method using spatially weighted LE and (4) the conventional MI-based method. Quantitative evaluations on 80 multimodal image pairs of different organs including 50 pairs of MR images with artificial deformations, 20 pairs of medical brain MR images and 10 pairs of breast images showed that our proposed method outperformed the comparison methods. Student's t test demonstrated that our method achieved statistically significant improvement on registration accuracy. The two-stage registration with joint ST and LE outperformed the conventional MI-based method for multimodal images. Both the ST and the LE contributed to the improved registration accuracy.

Minimally invasive registration for computer-assisted orthopedic surgery: combining tracked ultrasound and bone surface points via the P-IMLOP algorithm.

We present a registration method for computer-assisted total hip replacement (THR) surgery, which we demonstrate to improve the state of the art by both reducing the invasiveness of current methods and increasing registration accuracy. A critical element of computer-guided procedures is the determination of the spatial correspondence between the patient and a computational model of patient anatomy. The current method for establishing this correspondence in robot-assisted THR is to register points intraoperatively sampled by a tracked pointer from the exposed proximal femur and, via auxiliary incisions, from the distal femur. In this paper, we demonstrate a noninvasive technique for sampling points on the distal femur using tracked B-mode ultrasound imaging and present a new algorithm for registering these data called Projected Iterative Most-Likely Oriented Point (P-IMLOP). Points and normal orientations of the distal bone surface are segmented from ultrasound images and registered to the patient model along with points sampled from the exposed proximal femur via a tracked pointer. The proposed approach is evaluated using a bone- and tissue-mimicking leg phantom constructed to enable accurate assessment of experimental registration accuracy with respect to a CT-image-based model of the phantom. These experiments demonstrate that localization of the femur shaft is greatly improved by tracked ultrasound. The experiments further demonstrate that, for ultrasound-based data, the P-IMLOP algorithm significantly improves registration accuracy compared to the standard ICP algorithm. Registration via tracked ultrasound and the P-IMLOP algorithm has high potential to reduce the invasiveness and improve the registration accuracy of computer-assisted orthopedic procedures.

Using image synthesis for multi-channel registration of different image modalities.

This paper presents a multi-channel approach for performing registration between magnetic resonance (MR) images with different modalities. In general, a multi-channel registration cannot be used when the moving and target images do not have analogous modalities. In this work, we address this limitation by using a random forest regression technique to synthesize the missing modalities from the available ones. This allows a single channel registration between two different modalities to be converted into a multi-channel registration with two mono-modal channels. To validate our approach, two openly available registration algorithms and five cost functions were used to compare the label transfer accuracy of the registration with (and without) our multi-channel synthesis approach. Our results show that the proposed method produced statistically significant improvements in registration accuracy (at an α level of 0.001) for both algorithms and all cost functions when compared to a standard multi-modal registration using the same algorithms with mutual information.

Iterative most-likely point registration (IMLP): a robust algorithm for computing optimal shape alignment.

We present a probabilistic registration algorithm that robustly solves the problem of rigid-body alignment between two shapes with high accuracy, by aptly modeling measurement noise in each shape, whether isotropic or anisotropic. For point-cloud shapes, the probabilistic framework additionally enables modeling locally-linear surface regions in the vicinity of each point to further improve registration accuracy. The proposed Iterative Most-Likely Point (IMLP) algorithm is formed as a variant of the popular Iterative Closest Point (ICP) algorithm, which iterates between point-correspondence and point-registration steps. IMLP’s probabilistic framework is used to incorporate a generalized noise model into both the correspondence and the registration phases of the algorithm, hence its name as a most-likely point method rather than a closest-point method. To efficiently compute the most-likely correspondences, we devise a novel search strategy based on a principal direction (PD)-tree search. We also propose a new approach to solve the generalized total-least-squares (GTLS) sub-problem of the registration phase, wherein the point correspondences are registered under a generalized noise model. Our GTLS approach has improved accuracy, efficiency, and stability compared to prior methods presented for this problem and offers a straightforward implementation using standard least squares. We evaluate the performance of IMLP relative to a large number of prior algorithms including ICP, a robust variant on ICP, Generalized ICP (GICP), and Coherent Point Drift (CPD), as well as drawing close comparison with the prior anisotropic registration methods of GTLS-ICP and A-ICP. The performance of IMLP is shown to be superior with respect to these algorithms over a wide range of noise conditions, outliers, and misalignments using both mesh and point-cloud representations of various shapes.