Showing papers on "Image registration published in 2012"

Automatic camera and range sensor calibration using a single shot

Abstract: As a core robotic and vision problem, camera and range sensor calibration have been researched intensely over the last decades. However, robotic research efforts still often get heavily delayed by the requirement of setting up a calibrated system consisting of multiple cameras and range measurement units. With regard to removing this burden, we present a toolbox with web interface for fully automatic camera-to-camera and camera-to-range calibration. Our system is easy to setup and recovers intrinsic and extrinsic camera parameters as well as the transformation between cameras and range sensors within one minute. In contrast to existing calibration approaches, which often require user intervention, the proposed method is robust to varying imaging conditions, fully automatic, and easy to use since a single image and range scan proves sufficient for most calibration scenarios. Experimentally, we demonstrate that the proposed checkerboard corner detector significantly outperforms current state-of-the-art. Furthermore, the proposed camera-to-range registration method is able to discover multiple solutions in the case of ambiguities. Experiments using a variety of sensors such as grayscale and color cameras, the Kinect 3D sensor and the Velodyne HDL-64 laser scanner show the robustness of our method in different indoor and outdoor settings and under various lighting conditions.

Image Similarity and Tissue Overlaps as Surrogates for Image Registration Accuracy: Widely Used but Unreliable

Abstract: The accuracy of nonrigid image registrations is commonly approximated using surrogate measures such as tissue label overlap scores, image similarity, image difference, or transformation inverse consistency error. This paper provides experimental evidence that these measures, even when used in combination, cannot distinguish accurate from inaccurate registrations. To this end, we introduce a “registration” algorithm that generates highly inaccurate image transformations, yet performs extremely well in terms of the surrogate measures. Of the tested criteria, only overlap scores of localized anatomical regions reliably distinguish reasonable from inaccurate registrations, whereas image similarity and tissue overlap do not. We conclude that tissue overlap and image similarity, whether used alone or together, do not provide valid evidence for accurate registrations and should thus not be reported or accepted as such.

Motion correction in optical coherence tomography volumes on a per A-scan basis using orthogonal scan patterns

Abstract: High speed Optical Coherence Tomography (OCT) has made it possible to rapidly capture densely sampled 3D volume data. One key application is the acquisition of high quality in vivo volumetric data sets of the human retina. Since the volume is acquired in a few seconds, eye movement during the scan process leads to distortion, which limits the accuracy of quantitative measurements using 3D OCT data. In this paper, we present a novel software based method to correct motion artifacts in OCT raster scans. Motion compensation is performed retrospectively using image registration algorithms on the OCT data sets themselves. Multiple, successively acquired volume scans with orthogonal fast scan directions are registered retrospectively in order to estimate and correct eye motion. Registration is performed by optimizing a large scale numerical problem as given by a global objective function using one dense displacement field for each input volume and special regularization based on the time structure of the acquisition process. After optimization, each volume is undistorted and a single merged volume is constructed that has superior signal quality compared to the input volumes. Experiments were performed using 3D OCT data from the macula and optic nerve head acquired with a high-speed ultra-high resolution 850 nm spectral OCT as well as wide field data acquired with a 1050 nm swept source OCT instrument. Evaluation of registration performance and result stability as well as visual inspection shows that the algorithm can correct for motion in all three dimensions and on a per A-scan basis. Corrected volumes do not show visible motion artifacts. In addition, merging multiple motion corrected and registered volumes leads to improved signal quality. These results demonstrate that motion correction and merging improves image quality and should also improve morphometric measurement accuracy from volumetric OCT data.

MRI-Based Attenuation Correction for Hybrid PET/MRI Systems: A 4-Class Tissue Segmentation Technique Using a Combined Ultrashort-Echo-Time/Dixon MRI Sequence

Abstract: Accurate γ-photon attenuation correction (AC) is essential for quantitative PET/MRI as there is no simple relation between MR image intensity and attenuation coefficients. Attenuation maps (μ-maps) can be derived by segmenting MR images and assigning attenuation coefficients to the compartments. Ultrashort-echo-time (UTE) sequences have been used to separate cortical bone and air, and the Dixon technique has enabled differentiation between soft and adipose tissues. Unfortunately, sequential application of these sequences is time-consuming and complicates image registration. Methods: A UTE triple-echo (UTILE) MRI sequence is proposed, combining UTE sampling for bone detection and gradient echoes for Dixon water–fat separation in a radial 3-dimensional acquisition (repetition time, 4.1 ms; echo times, 0.09/1.09/2.09 ms; field strength, 3 T). Air masks are derived mainly from the phase information of the first echo; cortical bone is segmented using a dual-echo technique. Soft-tissue and adipose-tissue decomposition is achieved using a 3-point Dixon-like decomposition. Predefined linear attenuation coefficients are assigned to classified voxels to generate MRI-based μ-maps. The results of 6 patients are obtained by comparing μ-maps, reciprocal sensitivity maps, reconstructed PET images, and brain region PET activities based on either CT AC, two 3-class MRI AC techniques, or the proposed 4-class UTILE AC. Results: Using the UTILE MRI sequence, an acquisition time of 214 s was achieved for the head-and-neck region with 1.75-mm isotropic resolution, compared with 164 s for a single-echo UTE scan. MRI-based reciprocal sensitivity maps show a high correlation with those derived from CT scans (R2 = 0.9920). The same is true for PET activities (R2 = 0.9958). An overall voxel classification accuracy (compared with CT) of 81.1% was reached. Bone segmentation is inaccurate in complex regions such as the paranasal sinuses, but brain region activities in 48 regions across 6 patients show a high correlation after MRI-based and CT-based correction (R2 = 0.9956), with a regression line slope of 0.960. All overall correlations are higher and brain region PET activities more accurate in terms of mean and maximum deviations for the 4-class technique than for 3-class techniques. Conclusion: The UTILE MRI sequence enables the generation of MRI-based 4-class μ-maps without anatomic priors, yielding results more similar to CT-based results than can be obtained with 3-class segmentation only.

Image Registration: Principles, Tools and Methods

Abstract: This book presents a thorough and detailed guide to image registration, outlining the principles and reviewing state-of-the-art tools and methods. The book begins by identifying the components of a general image registration system, and then describes the design of each component using various image analysis tools. The text reviews a vast array of tools and methods, not only describing the principles behind each tool and method, but also measuring and comparing their performances using synthetic and real data. Features: discusses similarity/dissimilarity measures, point detectors, feature extraction/selection and homogeneous/heterogeneous descriptors; examines robust estimators, point pattern matching algorithms, transformation functions, and image resampling and blending; covers principal axes methods, hierarchical methods, optimization-based methods, edge-based methods, model-based methods, and adaptive methods; includes a glossary, an extensive list of references, and an appendix on PCA.

GLISTR: Glioma Image Segmentation and Registration

Abstract: We present a generative approach for simultaneously registering a probabilistic atlas of a healthy population to brain magnetic resonance (MR) scans showing glioma and segmenting the scans into tumor as well as healthy tissue labels. The proposed method is based on the expectation maximization (EM) algorithm that incorporates a glioma growth model for atlas seeding, a process which modifies the original atlas into one with tumor and edema adapted to best match a given set of patient's images. The modified atlas is registered into the patient space and utilized for estimating the posterior probabilities of various tissue labels. EM iteratively refines the estimates of the posterior probabilities of tissue labels, the deformation field and the tumor growth model parameters. Hence, in addition to segmentation, the proposed method results in atlas registration and a low-dimensional description of the patient scans through estimation of tumor model parameters. We validate the method by automatically segmenting 10 MR scans and comparing the results to those produced by clinical experts and two state-of-the-art methods. The resulting segmentations of tumor and edema outperform the results of the reference methods, and achieve a similar accuracy from a second human rater. We additionally apply the method to 122 patients scans and report the estimated tumor model parameters and their relations with segmentation and registration results. Based on the results from this patient population, we construct a statistical atlas of the glioma by inverting the estimated deformation fields to warp the tumor segmentations of patients scans into a common space.

Motion correction for myocardial T1 mapping using image registration with synthetic image estimation.

Abstract: Quantification of myocardial T1 relaxation has potential value in the diagnosis of both ischemic and nonischemic cardiomyopathies. Image acquisition using the modified Look-Locker inversion recovery technique is clinically feasible for T1 mapping. However, respiratory motion limits its applicability and degrades the accuracy of T1 estimation. The robust registration of acquired inversion recovery images is particularly challenging due to the large changes in image contrast, especially for those images acquired near the signal null point of the inversion recovery and other inversion times for which there is little tissue contrast. In this article, we propose a novel motion correction algorithm. This approach is based on estimating synthetic images presenting contrast changes similar to the acquired images. The estimation of synthetic images is formulated as a variational energy minimization problem. Validation on a consecutive patient data cohort shows that this strategy can perform robust nonrigid registration to align inversion recovery images experiencing significant motion and lead to suppression of motion induced artifacts in the T1 map.

MRI-Based Nonrigid Motion Correction in Simultaneous PET/MRI

Abstract: Respiratory and cardiac motion is the most serious limitation to whole-body PET, resulting in spatial resolution close to 1 cm. Furthermore, motion-induced inconsistencies in the attenuation measurements often lead to significant artifacts in the reconstructed images. Gating can remove motion artifacts at the cost of increased noise. This paper presents an approach to respiratory motion correction using simultaneous PET/MRI to demonstrate initial results in phantoms, rabbits, and nonhuman primates and discusses the prospects for clinical application. Methods: Studies with a deformable phantom, a free-breathing primate, and rabbits implanted with radioactive beads were performed with simultaneous PET/MRI. Motion fields were estimated from concurrently acquired tagged MR images using 2 B-spline nonrigid image registration methods and incorporated into a PET list-mode ordered-subsets expectation maximization algorithm. Using the measured motion fields to transform both the emission data and the attenuation data, we could use all the coincidence data to reconstruct any phase of the respiratory cycle. We compared the resulting SNR and the channelized Hotelling observer (CHO) detection signal-to-noise ratio (SNR) in the motion-corrected reconstruction with the results obtained from standard gating and uncorrected studies. Results: Motion correction virtually eliminated motion blur without reducing SNR, yielding images with SNR comparable to those obtained by gating with 5–8 times longer acquisitions in all studies. The CHO study in dynamic phantoms demonstrated a significant improvement (166%–276%) in lesion detection SNR with MRI-based motion correction as compared with gating (P

Scale-Invariant Features for 3-D Mesh Models

Abstract: In this paper, we present a framework for detecting interest points in 3-D meshes and computing their corresponding descriptors. For that, we propose an intrinsic scale detection scheme per interest point and utilize it to derive two scale-invariant local features for mesh models. First, we present the scale-invariant spin image local descriptor that is a scale-invariant formulation of the spin image descriptor. Second, we adapt the scale-invariant feature transform feature to mesh data by representing the vicinity of each interest point as a depth map and estimating its dominant angle using the principal component analysis to achieve rotation invariance. The proposed features were experimentally shown to be robust to scale changes and partial mesh matching, and they were compared favorably with other local mesh features on the SHREC'10 and SHREC'11 testbeds. We applied the proposed local features to mesh retrieval using the bag-of-features approach and achieved state-of-the-art retrieval accuracy. Last, we applied the proposed local features to register models to scanned depth scenes and achieved high registration accuracy.

The need for application-based adaptation of deformable image registration.

Abstract: Purpose: To utilize a deformable phantom to objectively evaluate the accuracy of 11 different deformable image registration (DIR) algorithms. Methods: The phantom represents an axial plane of the pelvic anatomy. Urethane plastic serves as the bony anatomy and urethane rubber with three levels of Hounsfield units (HU) is used to represent fat and organs, including the prostate. A plastic insert is placed into the phantom to simulate bladder filling. Nonradiopaque markers reside on the phantom surface. Optical camera images of these markers are used to measure the positions and determine the deformation from the bladder insert. Eleven different DIR algorithms are applied to the full and empty-bladder computed tomography images of the phantom (fixed and moving volumes, respectively) to calculate the deformation. The algorithms include those fromMIM Software (MIM) and Velocity Medical Solutions (VEL) and nine different implementations from the deformable image registration and adaptive radiotherapy toolbox for Matlab. These algorithms warp one image to make it similar to another, but must utilize a method for regularization to avoid physically unrealistic deformation scenarios. The mean absolute difference (MAD) between the HUs at the marker locations on one image and the calculated location on the other serves as a metric to evaluate the balance between image similarity and regularization. To demonstrate the effect of regularization on registration accuracy, an additional beta version of MIM was created with a variable smoothness factor that controls the emphasis of the algorithm on regularization. The distance to agreement between the measured and calculated marker deformations is used to compare the overall spatial accuracy of the DIR algorithms. This overall spatial accuracy is also utilized to evaluate the phantom geometry and the ability of the phantom soft-tissue heterogeneity to represent patient data. To evaluate the ability of the DIR algorithms to accurately transfer anatomical contours, the rectum is delineated on both the fixed and moving images. A Dice similarity coefficient is then calculated between the contour on the fixed image and that transferred, via the calculated deformation, from the moving to the fixed image. Results: The phantom possesses sufficient soft-tissue heterogeneity to act as a proxy for patient data. Large discrepancies appear between the algorithms and the measured ground-truth deformation. VEL yields the smallest mean spatial error and a Dice coefficient of 0.90. MIM produces the lowest MAD value and the highest Dice coefficient of 0.96, but creates the largest spatial errors. Increasing theMIM smoothness factor above the default value improves the overall spatial accuracy, but the factor associated with the lowest mean error decreases the Dice coefficient to 0.85. Conclusions: Different applications of DIR require disparate balances between image similarity and regularization. A DIR algorithm that is optimized only for its ability to transfer anatomical contours will yield large deformation errors in homogeneous regions, which is problematic for dose mapping. For this reason, these algorithms must be tested for their overall spatial accuracy. The developed phantom is an objective tool for this purpose.

BFSIFT: A Novel Method to Find Feature Matches for SAR Image Registration

Abstract: In this letter, we propose a novel method based on bilateral filter (BF) scale-invariant feature transform (SIFT) (BFSIFT) to find feature matches for synthetic aperture radar (SAR) image registration. First, the anisotropic scale space of the image is constructed using BFs. The constructing process is noniterative and fast. Compared with the Gaussian scale space used in SIFT, more accurately located matches can be found in the anisotropic one. Then, keypoints are detected and described in the coarser scales using SIFT. At last, dual-matching strategy and random sample consensus are used to establish matches. The probability of correct matching is significantly increased by skipping the finest scale and by the dual-matching strategy. Experiments on various slant range images demonstrate the applicability of BFSIFT to find feature matches for SAR image registration.

How to Compare Noisy Patches? Patch Similarity Beyond Gaussian Noise

Abstract: Many tasks in computer vision require to match image parts. While higher-level methods consider image features such as edges or robust descriptors, low-level approaches (so-called image-based) compare groups of pixels (patches) and provide dense matching. Patch similarity is a key ingredient to many techniques for image registration, stereo-vision, change detection or denoising. Recent progress in natural image modeling also makes intensive use of patch comparison. A fundamental difficulty when comparing two patches from "real" data is to decide whether the differences should be ascribed to noise or intrinsic dissimilarity. Gaussian noise assumption leads to the classical definition of patch similarity based on the squared differences of intensities. For the case where noise departs from the Gaussian distribution, several similarity criteria have been proposed in the literature of image processing, detection theory and machine learning. By expressing patch (dis)similarity as a detection test under a given noise model, we introduce these criteria with a new one and discuss their properties. We then assess their performance for different tasks: patch discrimination, image denoising, stereo-matching and motion-tracking under gamma and Poisson noises. The proposed criterion based on the generalized likelihood ratio is shown to be both easy to derive and powerful in these diverse applications.

Motion Correction in Dual Gated Cardiac PET Using Mass-Preserving Image Registration

Abstract: Respiratory and cardiac motion leads to image degradation in positron emission tomography (PET) studies of the human heart. In this paper we present a novel approach to motion correction based on dual gating and mass-preserving hyperelastic image registration. Thereby, we account for intensity modulations caused by the highly nonrigid cardiac motion. This leads to accurate and realistic motion estimates which are quantitatively validated on software phantom data and carried over to clinically relevant data using a hardware phantom. For patient data, the proposed method is first evaluated in a high statistic (20 min scans) dual gating study of 21 patients. It is shown that the proposed approach properly corrects PET images for dual-cardiac as well as respiratory-motion. In a second study the list mode data of the same patients is cropped to a scan time reasonable for clinical practice (3 min). This low statistic study not only shows the clinical applicability of our method but also demonstrates its robustness against noise obtained by hyperelastic regularization.

Toward Routine Use of 3D Histopathology as a Research Tool

Abstract: Three-dimensional (3D) reconstruction and examination of tissue at microscopic resolution have significant potential to enhance the study of both normal and disease processes, particularly those involving structural changes or those in which the spatial relationship of disease features is important. Although other methods exist for studying tissue in 3D, using conventional histopathological features has significant advantages because it allows for conventional histopathological staining and interpretation techniques. Until now, its use has not been routine in research because of the technical difficulty in constructing 3D tissue models. We describe a novel system for 3D histological reconstruction, integrating whole-slide imaging (virtual slides), image serving, registration, and visualization into one user-friendly package. It produces high-resolution 3D reconstructions with minimal user interaction and can be used in a histopathological laboratory without input from computing specialists. It uses a novel method for slice-to-slice image registration using automatic registration algorithms custom designed for both virtual slides and histopathological images. This system has been applied to >300 separate 3D volumes from eight different tissue types, using a total of 5500 virtual slides comprising 1.45 TB of primary image data. Qualitative and quantitative metrics for the accuracy of 3D reconstruction are provided, with measured registration accuracy approaching 120 μm for a 1-cm piece of tissue. Both 3D tissue volumes and generated 3D models are presented for four demonstrator cases.

Online robust image alignment via iterative convex optimization

Abstract: In this paper we study the problem of online aligning a newly arrived image to previously well-aligned images. Inspired by recent advances in batch image alignment using low rank decomposition [ ], we treat the newly arrived image, after alignment, as being linearly and sparsely reconstructed by the well-aligned ones. The task is accomplished by a sequence of convex optimization that minimizes the l\-norm. After that, online basis updating is pursued in two different ways: (1) a two-stage incremental alignment for joint registration of a large image dataset which is known a prior, and (2) a greedy online alignment of dynamically increasing image sequences, such as in the tracking scenario. In (1), we first sequentially collect basis images that are easily aligned by checking their reconstruction residuals, followed by the second stage where all images are re-aligned one-by-one using the collected basis set. In (2), during the tracking process, we dynamically enrich the image basis set by the new target if it significantly distinguishes itself from existing basis images. While inheriting the benefits of sparsity, our method enjoys the great time efficiency and therefore be capable of dealing with large image set and real time tasks such as visual tracking. The efficacy of the proposed online robust alignment algorithm is verified with extensive experiments on image set alignment and visual tracking, in reference with state-of-the-art methods.

Convergent Iterative Closest-Point Algorithm to Accomodate Anisotropic and Inhomogenous Localization Error

Abstract: Since its introduction in the early 1990s, the Iterative Closest Point (ICP) algorithm has become one of the most well-known methods for geometric alignment of 3D models. Given two roughly aligned shapes represented by two point sets, the algorithm iteratively establishes point correspondences given the current alignment of the data and computes a rigid transformation accordingly. From a statistical point of view, however, it implicitly assumes that the points are observed with isotropic Gaussian noise. In this paper, we show that this assumption may lead to errors and generalize the ICP such that it can account for anisotropic and inhomogenous localization errors. We 1) provide a formal description of the algorithm, 2) extend it to registration of partially overlapping surfaces, 3) prove its convergence, 4) derive the required covariance matrices for a set of selected applications, and 5) present means for optimizing the runtime. An evaluation on publicly available surface meshes as well as on a set of meshes extracted from medical imaging data shows a dramatic increase in accuracy compared to the original ICP, especially in the case of partial surface registration. As point-based surface registration is a central component in various applications, the potential impact of the proposed method is high.

Intraoperative Image-based Multiview 2D/3D Registration for Image-Guided Orthopaedic Surgery: Incorporation of Fiducial-Based C-Arm Tracking and GPU-Acceleration

Abstract: Intraoperative patient registration may significantly affect the outcome of image-guided surgery (IGS). Image-based registration approaches have several advantages over the currently dominant point-based direct contact methods and are used in some industry solutions in image-guided radiation therapy with fixed X-ray gantries. However, technical challenges including geometric calibration and computational cost have precluded their use with mobile C-arms for IGS. We propose a 2D/3D registration framework for intraoperative patient registration using a conventional mobile X-ray imager combining fiducial-based C-arm tracking and graphics processing unit (GPU)-acceleration. The two-stage framework 1) acquires X-ray images and estimates relative pose between the images using a custom-made in-image fiducial, and 2) estimates the patient pose using intensity-based 2D/3D registration. Experimental validations using a publicly available gold standard dataset, a plastic bone phantom and cadaveric specimens have been conducted. The mean target registration error (mTRE) was 0.34±0.04 mm (success rate: 100%, registration time: 14.2 s) for the phantom with two images 90° apart, and 0.99±0.41 mm (81%, 16.3 s) for the cadaveric specimen with images 58.5° apart. The experimental results showed the feasibility of the proposed registration framework as a practical alternative for IGS routines.

Multifeature Landmark-Free Active Appearance Models: Application to Prostate MRI Segmentation

Abstract: Active shape models (ASMs) and active appearance models (AAMs) are popular approaches for medical image segmentation that use shape information to drive the segmentation process. Both approaches rely on image derived landmarks (specified either manually or automatically) to define the object's shape, which require accurate triangulation and alignment. An alternative approach to modeling shape is the level-set representation, defined as a set of signed distances to the object's surface. In addition, using multiple image derived attributes (IDAs) such as gradient information has previously shown to offer improved segmentation results when applied to ASMs, yet little work has been done exploring IDAs in the context of AAMs. In this work, we present a novel AAM methodology that utilizes the level set implementation to overcome the issues relating to specifying landmarks, and locates the object of interest in a new image using a registration based scheme. Additionally, the framework allows for incorporation of multiple IDAs. Our multifeature landmark-free AAM (MFLAAM) utilizes an efficient, intuitive, and accurate algorithm for identifying those IDAs that will offer the most accurate segmentations. In this paper, we evaluate our MFLAAM scheme for the problem of prostate segmentation from T2-w MRI volumes. On a cohort of 108 studies, the levelset MFLAAM yielded a mean Dice accuracy of 88% ± 5%, and a mean surface error of 1.5 mm ± .8 mm with a segmentation time of 150/s per volume. In comparison, a state of the art AAM yielded mean Dice and surface error values of 86% ± 9% and 1.6 mm ± 1.0 mm, respectively. The differences with respect to our levelset-based MFLAAM model are statistically significant (p <; .05). In addition, our results were in most cases superior to several recent state of the art prostate MRI segmentation methods.

A Simple and Robust Feature Point Matching Algorithm Based on Restricted Spatial Order Constraints for Aerial Image Registration

Abstract: Accurate point matching is a critical and challenging process in feature-based image registration. In this paper, a simple and robust feature point matching algorithm, called Restricted Spatial Order Constraints (RSOC), is proposed to remove outliers for registering aerial images with monotonous backgrounds, similar patterns, low overlapping areas, and large affine transformation. In RSOC, both local structure and global information are considered. Based on adjacent spatial order, an affine invariant descriptor is defined, and point matching is formulated as an optimization problem. A graph matching method is used to solve it and yields two matched graphs with a minimum global transformation error. In order to eliminate dubious matches, a filtering strategy is designed. The strategy integrates two-way spatial order constraints and two decision criteria restrictions, i.e., the stability and accuracy of transformation error. Twenty-nine pairs of optical and Synthetic Aperture Radar (SAR) aerial images are utilized to evaluate the performance. Compared with RANdom SAmple Consensus (RANSAC), Graph Transformation Matching (GTM), and Spatial Order Constraints (SOC), RSOC obtained the highest precision and stability.

Feature-Based Deformable Surface Detection with Self-Occlusion Reasoning

Abstract: This paper presents a method for detecting a textured deformed surface in an image. It uses (wide-baseline) point matches between a template and the input image. The main contribution of the paper is twofold. First, we propose a robust method based on local surface smoothness capable of discarding outliers from the set of point matches. Our method handles large proportions of outliers (beyond 70% with less than 15% of false positives) even when the surface self-occludes. Second, we propose a method to estimate a self-occlusion resistant warp from point matches. Our method allows us to realistically retexture the input image. A pixel-based (direct) registration approach is also proposed. Bootstrapped by our robust point-based method, it finely tunes the warp parameters using the value (intensity or color) of all the visible surface pixels. The proposed framework was tested with simulated and real data. Convincing results are shown for the detection and retexturing of deformed surfaces in challenging images.

Diffeomorphic susceptibility artifact correction of diffusion-weighted magnetic resonance images

Abstract: Diffusion-weighted magnetic resonance imaging is a key investigation technique in modern neuroscience. In clinical settings, diffusion-weighted imaging and its extension to diffusion tensor imaging (DTI) are usually performed applying the technique of echo-planar imaging (EPI). EPI is the commonly available ultrafast acquisition technique for single-shot acquisition with spatial encoding in a Cartesian system. A drawback of these sequences is their high sensitivity against small perturbations of the magnetic field, caused, e.g., by differences in magnetic susceptibility of soft tissue, bone and air. The resulting magnetic field inhomogeneities thus cause geometrical distortions and intensity modulations in diffusion-weighted images. This complicates the fusion with anatomical T1- or T2-weighted MR images obtained with conventional spin- or gradient-echo images and negligible distortion. In order to limit the degradation of diffusion-weighted MR data, we present here a variational approach based on a reference scan pair with reversed polarity of the phase- and frequency-encoding gradients and hence reversed distortion. The key novelty is a tailored nonlinear regularization functional to obtain smooth and diffeomorphic transformations. We incorporate the physical distortion model into a variational image registration framework and derive an accurate and fast correction algorithm. We evaluate the applicability of our approach to distorted DTI brain scans of six healthy volunteers. For all datasets, the automatic correction algorithm considerably reduced the image degradation. We show that, after correction, fusion with T1- or T2-weighted images can be obtained by a simple rigid registration. Furthermore, we demonstrate the improvement due to the novel regularization scheme. Most importantly, we show that it provides meaningful, i.e. diffeomorphic, geometric transformations, independent of the actual choice of the regularization parameters.

Dynamic facial expression recognition using longitudinal facial expression atlases

Abstract: In this paper, we propose a new scheme to formulate the dynamic facial expression recognition problem as a longitudinal atlases construction and deformable groupwise image registration problem. The main contributions of this method include: 1) We model human facial feature changes during the facial expression process by a diffeomorphic image registration framework; 2) The subject-specific longitudinal change information of each facial expression is captured by building an expression growth model; 3) Longitudinal atlases of each facial expression are constructed by performing groupwise registration among all the corresponding expression image sequences of different subjects. The constructed atlases can reflect overall facial feature changes of each expression among the population, and can suppress the bias due to inter-personal variations. The proposed method was extensively evaluated on the Cohn-Kanade, MMI, and Oulu-CASIA VIS dynamic facial expression databases and was compared with several state-of-the-art facial expression recognition approaches. Experimental results demonstrate that our method consistently achieves the highest recognition accuracies among other methods under comparison on all the databases.

Development of patient-specific biomechanical models for predicting large breast deformation.

Abstract: Physically realistic simulations for large breast deformation are of great interest for many medical applications such as cancer diagnosis, image registration, surgical planning and image-guided surgery. To support fast, large deformation simulations of breasts in clinical settings, we proposed a patient-specific biomechanical modelling framework for breasts, based on an open-source graphics processing unit-based, explicit, dynamic, nonlinear finite element (FE) solver. A semi-automatic segmentation method for tissue classification, integrated with a fully automated FE mesh generation approach, was implemented for quick patient-specific FE model generation. To solve the difficulty in determining material parameters of soft tissues in vivo for FE simulations, a novel method for breast modelling, with a simultaneous material model parameter optimization for soft tissues in vivo, was also proposed. The optimized deformation prediction was obtained through iteratively updating material model parameters to maximize the image similarity between the FE-predicted MR image and the experimentally acquired MR image of a breast. The proposed method was validated and tested by simulating and analysing breast deformation experiments under plate compression. Its prediction accuracy was evaluated by calculating landmark displacement errors. The results showed that both the heterogeneity and the anisotropy of soft tissues were essential in predicting large breast deformations under plate compression. As a generalized method, the proposed process can be used for fast deformation analyses of soft tissues in medical image analyses and surgical simulations.

Simultaneous PET/MR imaging: MR-based attenuation correction of local radiofrequency surface coils.

Abstract: Purpose: In simultaneous positron emission tomography/magnetic resonance (PET/MR) imaging, local receiver surface radiofrequency (RF) coils are positioned in the field-of-view (FOV) of the PETdetector during PET/MR data acquisition and potentially attenuate the PET signal. For flexible body RF surface coils placed on top of the patient's body, MR-based attenuation correction (AC) is an unsolved problem since the RF coils are not inherently visible in MRimages and their individual position in the FOV is patient specific and not knowna priori. The aim of this work was to quantify the effect of local body RF coils used in the Biograph mMR hybrid PET/MR system on PET emission data and to present techniques for MR-based position determination of these specific local RF coils. Methods: Acquisitions of a homogeneous phantom were performed on a whole-body PET/MRI scanner. Two different PET emission scans were performed, with and without the local body matrix RF coil placed on the top of the phantom. For position determination of the coil, two methods were applied. First, cod liver oil capsules were attached to the surface of the coil and second, an ultrashort echo time (UTE) sequence was used. PETimages were reconstructed in five different ways: (1) PET reference scan without the coil, (2) PET scan with the coil, but omitting the coil in AC (PET/MR scanning conditions), (3) AC of the coil using a CT scan of the same phantom setup and registration via capsules, (4) same setup as 3, but registration was done using UTE images, neglecting the capsules, and (5) registration using the capsules, but the CT was performed with the coil placed flat on the CT table and the outer regions of the coil were cropped. The activity concentrations were then compared to the reference scan. For clinical evaluation of the concept, the presented methods were also evaluated on a patient. Results: The oil capsules were visible in the MR and CTimages and image registration was straightforward. The UTE images show only parts of the coil's plastic housing and image registration was more difficult. The overall loss of true counts due to the presence of the surface coil is 4.7%. However, a spatially dependent analysis shows larger deviation (10%–15% attenuation) of the activity concentration in the top part of the phantom close to the coil. When accounting for the RF coil for PET AC, attenuation due to the RF coil could mostly be corrected. These results of the phantom studies were confirmed by the patient measurements. Conclusions: Disregarding local coils in PET AC can lead to a bias of the AC PETimages that is regional dependent. The closer the analyzed region is located to the coil, the higher the bias. Cod liver oil capsules or the UTE sequence can be used for RF coil position determination. The middle part of the examined RF coil hosting the preamplifiers and electronic components provides the highest attenuating part. Consequently, emphasis should be put on correcting for this portion of the RF coils with the suggested methods.