Attribute vector guided groupwise registration

An artificial-intelligence-based age-specific template construction framework for brain structural analysis using magnetic resonance images

It is an essential task to construct brain templates and analyze their anatomical structures in neurological and cognitive science. Generally, templates constructed from magnetic resonance imaging (MRI) of a group of subjects can provide a standard reference space for analyzing the structural and functional characteristics of the group. With recent development of artificial intelligence (AI) techniques, it is desirable to explore AI registration methods for quantifying age-specific brain variations and tendencies across different ages. In this article, we present an AI-based age-specific template construction (called ASTC) framework for longitudinal structural brain analysis using T1-weighted MRIs of 646 subjects from 18 to 82 years old collected from four medical centers. Altogether, 13 longitudinal templates were constructed at a 5-year age interval using ASTC, and tissue segmentation and substructure parcellation were performed for analysis across different age groups. The results indicated consistent changes in brain structures along with aging and demonstrated the capability of ASTC for longitudinal neuroimaging study.

Reference-free brain template construction with population symmetric registration

Population registration has been proposed for normalizing a large group of images into a common space, which is important in many clinical and research studies, such as brain development, aging, and atlas construction. Different from pairwise registration problem that aligns the target image to the reference directly, determining the reference or the hidden common space with the least bias is important in population registration. In order to decrease this bias, a lot of work takes the arithmetic mean image as the reference. However, the arithmetic mean image is usually too smooth to guide the population registration. This work presents an efficient symmetric population registration strategy for brain template construction, which defines the symmetric population center guiding population registration. This is important because the population registration problem can be translated into a series of pairwise registration problem which is easier to optimize and implement. Another prominent merit of proposed population registration algorithm is reference-free, which eliminates the reference dependency-related bias in population registration. Based on this symmetric population registration, the brain template is constructed by approximating both the population's intensity and gradient information. In addition, we also present a new measurement named with average bias for evaluating the unbiasedness of brain template. Experiments were first carried out on four synthetic images created with controllable transforms, which aim at comparing the difference between conventional method and proposed method. Further experiment is designed for reference-free validation. Finally, in real inter-subject brain data, twenty MRI T1 volumes with size 256 × 256 × 176 are used to construct a symmetric brain template with proposed population registration method. The constructed brain template has a small bias and clear brain details comparing with DARTEL.

Fast Groupwise Registration Using Multi-Level and Multi-Resolution Graph Shrinkage

Groupwise registration aligns a set of images to a common space. It can however be inefficient and ineffective when dealing with datasets with significant anatomical variations. To mitigate these problems, we propose a groupwise registration framework based on hierarchical multi-level and multi-resolution shrinkage of a graph set. First, to deal with datasets with complex inhomogeneous image distributions, we divide the images hierarchically into multiple clusters. Since the images in each cluster have similar appearances, they can be registered effectively. Second, we employ a multi-resolution strategy to reduce computational cost. Experimental results on two public datasets show that our proposed method yields state-of-the-art registration accuracy with significantly reduced computational time.

Spatially Consistent Supervoxel Correspondences of Cone-Beam Computed Tomography Images

Establishing dense correspondences of cone-beam computed tomography (CBCT) images is a crucial step for the attribute transfer and morphological variation assessment in clinical orthodontics. In this paper, a novel method, unsupervised spatially consistent clustering forest, is proposed to tackle the challenges for automatic supervoxel-wise correspondences of CBCT images. A complexity analysis of the proposed method with respect to the clustering hypotheses is provided with a data-dependent learning guarantee. The learning bound considers both the sequential tree traversals determined by questions stored in branch nodes and the clustering compactness of leaf nodes. A novel tree-pruning algorithm, guided by the learning bound, is also proposed to remove locally inconsistent leaf nodes. The resulting forest yields spatially consistent affinity estimations, thanks to the pruning penalizing trees with inconsistent leaf assignments and the combinational contextual feature channels used to learn the forest. A forest-based metric is utilized to derive the pairwise affinities and dense correspondences of CBCT images. The proposed method has been applied to the label propagation of clinically captured CBCT images. In the experiments, the method outperforms variants of both supervised and unsupervised forest-based methods and state-of-the-art label-propagation methods, achieving the mean dice similarity coefficients of 0.92, 0.89, 0.94, and 0.93 for the mandible, the maxilla, the zygoma arch, and the teeth data, respectively.

Concatenated Spatially-localized Random Forests for Hippocampus Labeling in Adult and Infant MR Brain Images

Automatic labeling of the hippocampus in brain MR images is highly demanded, as it has played an important role in imaging-based brain studies. However, accurate labeling of the hippocampus is still challenging, partially due to the ambiguous intensity boundary between the hippocampus and surrounding anatomies. In this paper, we propose a concatenated set of spatially-localized random forests for multi-atlas-based hippocampus labeling of adult/infant brain MR images. The contribution in our work is two-fold. First, each forest classifier is trained to label just a specific sub-region of the hippocampus, thus enhancing the labeling accuracy. Second, a novel forest selection strategy is proposed, such that each voxel in the test image can automatically select a set of optimal forests, and then dynamically fuses their respective outputs for determining the final label. Furthermore, we enhance the spatially-localized random forests with the aid of the auto-context strategy. In this way, our proposed learning framework can gradually refine the tentative labeling result for better performance. Experiments show that, regarding the large datasets of both adult and infant brain MR images, our method owns satisfactory scalability by segmenting the hippocampus accurately and efficiently.

Brain Atlas Fusion from High-Thickness Diagnostic Magnetic Resonance Images by Learning-Based Super-Resolution

It is fundamentally important to fuse the brain atlas from magnetic resonance (MR) images for many imaging-based studies. Most existing works focus on fusing the atlases from high-quality MR images. However, for low-quality diagnostic images (i.e., with high inter-slice thickness), the problem of atlas fusion has not been addressed yet. In this paper, we intend to fuse the brain atlas from the high-thickness diagnostic MR images that are prevalent for clinical routines. The main idea of our works is to extend the conventional groupwise registration by incorporating a novel super-resolution strategy. The contribution of the proposed super-resolution framework is two-fold. First, each high-thickness subject image is reconstructed to be isotropic by the patch-based sparsity learning. Then, the reconstructed isotropic image is enhanced for better quality through the random-forest-based regression model. In this way, the images obtained by the super-resolution strategy can be fused together by applying the groupwise registration method to construct the required atlas. Our experiments have shown that the proposed framework can effectively solve the problem of atlas fusion from the low-quality brain MR images.

Scalable High-Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning

Feature selection is a critical step in deformable image registration. In particular, selecting the most discriminative features that accurately and concisely describe complex morphological patterns in image patches improves correspondence detection, which in turn improves image registration accuracy. Furthermore, since more and more imaging modalities are being invented to better identify morphological changes in medical imaging data, the development of deformable image registration method that scales well to new image modalities or new image applications with little to no human intervention would have a significant impact on the medical image analysis community. To address these concerns, a learning-based image registration framework is proposed that uses deep learning to discover compact and highly discriminative features upon observed imaging data. Specifically, the proposed feature selection method uses a convolutional stacked autoencoder to identify intrinsic deep feature representations in image patches. Since deep learning is an unsupervised learning method, no ground truth label knowledge is required. This makes the proposed feature selection method more flexible to new imaging modalities since feature representations can be directly learned from the observed imaging data in a very short amount of time. Using the LONI and ADNI imaging datasets, image registration performance was compared to two existing state-of-the-art deformable image registration methods that use handcrafted features. To demonstrate the scalability of the proposed image registration framework, image registration experiments were conducted on 7.0-T brain MR images. In all experiments, the results showed that the new image registration framework consistently demonstrated more accurate registration results when compared to state of the art.

Scalable High-Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning

Feature selection is a critical step in deformable image registration. In particular, selecting the most discriminative features that accurately and concisely describe complex morphological patterns in image patches improves correspondence detection, which in turn improves image registration accuracy. Furthermore, since more and more imaging modalities are being invented to better identify morphological changes in medical imaging data, the development of deformable image registration method that scales well to new image modalities or new image applications with little to no human intervention would have a significant impact on the medical image analysis community. To address these concerns, a learning-based image registration framework is proposed that uses deep learning to discover compact and highly discriminative features upon observed imaging data. Specifically, the proposed feature selection method uses a convolutional stacked autoencoder to identify intrinsic deep feature representations in image patches. Since deep learning is an unsupervised learning method, no ground truth label knowledge is required. This makes the proposed feature selection method more flexible to new imaging modalities since feature representations can be directly learned from the observed imaging data in a very short amount of time. Using the LONI and ADNI imaging datasets, image registration performance was compared to two existing state-of-the-art deformable image registration methods that use handcrafted features. To demonstrate the scalability of the proposed image registration framework, image registration experiments were conducted on 7.0-T brain MR images. In all experiments, the results showed that the new image registration framework consistently demonstrated more accurate registration results when compared to state of the art.

Improved image registration by sparse patch-based deformation estimation

Despite intensive efforts for decades, deformable image registration is still a challenging problem due to the potential large anatomical differences across individual images, which limits the registration performance. Fortunately, this issue could be alleviated if a good initial deformation can be provided for the two images under registration, which are often termed as the moving subject and the fixed template, respectively. In this work, we present a novel patch-based initial deformation prediction framework for improving the performance of existing registration algorithms. Our main idea is to estimate the initial deformation between subject and template in a patch-wise fashion by using the sparse representation technique. We argue that two image patches should follow the same deformation toward the template image if their patch-wise appearance patterns are similar. To this end, our framework consists of two stages, i.e., the training stage and the application stage. In the training stage, we register all training images to the pre-selected template, such that the deformation of each training image with respect to the template is known. In the application stage, we apply the following four steps to efficiently calculate the initial deformation field for the new test subject: (1) We pick a small number of key points in the distinctive regions of the test subject; (2) for each key point, we extract a local patch and form a coupled appearance-deformation dictionary from training images where each dictionary atom consists of the image intensity patch as well as their respective local deformations; (3) a small set of training image patches in the coupled dictionary are selected to represent the image patch of each subject key point by sparse representation. Then, we can predict the initial deformation for each subject key point by propagating the pre-estimated deformations on the selected training patches with the same sparse representation coefficients; and (4) we employ thin-plate splines (TPS) to interpolate a dense initial deformation field by considering all key points as the control points. Thus, the conventional image registration problem becomes much easier in the sense that we only need to compute the remaining small deformation for completing the registration of the subject to the template. Experimental results on both simulated and real data show that the registration performance can be significantly improved after integrating our patch-based deformation prediction framework into the existing registration algorithms.

Intensity and sulci landmark combined brain atlas construction for Chinese pediatric population

Constructing an atlas from a population of brain images is of vital importance to medical image analysis. Especially in neuroscience study, creating a brain atlas is useful for intra- and inter-population comparison. Research on brain atlas construction has attracted great attention in recent years, but the research on pediatric population is still limited, mainly due to the limited availability and the relatively low quality of pediatric magnetic resonance brain images. This article is targeted at creating a high quality representative brain atlas for Chinese pediatric population. To achieve this goal, we have designed a set of preprocessing procedures to improve the image quality and developed an intensity and sulci landmark combined groupwise registration method to align the population of images for atlas construction. As demonstrated in experiments, the newly constructed atlas can better represent the size and shape of brains of Chinese pediatric population, and show better performance in Chinese pediatric brain image analysis compared with other standard atlases.

Diffusion tensor image registration using hybrid connectivity and tensor features

Most existing diffusion tensor imaging (DTI) registration methods estimate structural correspondences based on voxelwise matching of tensors. The rich connectivity information that is given by DTI, however, is often neglected. In this article, we propose to integrate complementary information given by connectivity features and tensor features for improved registration accuracy. To utilize connectivity information, we place multiple anchors representing different brain anatomies in the image space, and define the connectivity features for each voxel as the geodesic distances from all anchors to the voxel under consideration. The geodesic distance, which is computed in relation to the tensor field, encapsulates information of brain connectivity. We also extract tensor features for every voxel to reflect the local statistics of tensors in its neighborhood. We then combine both connectivity features and tensor features for registration of tensor images. From the images, landmarks are selected automatically and their correspondences are determined based on their connectivity and tensor feature vectors. The deformation field that deforms one tensor image to the other is iteratively estimated and optimized according to the landmarks and their associated correspondences. Experimental results show that, by using connectivity features and tensor features simultaneously, registration accuracy is increased substantially compared with the cases using either type of features alone.

Registration of challenging pre-clinical brain images

The size and complexity of brain imaging studies in pre-clinical populations are increasing, and automated image analysis pipelines are urgently required. Pre-clinical populations can be subjected to controlled interventions (e.g., targeted lesions), which significantly change the appearance of the brain obtained by imaging. Existing systems for registration (the systematic alignment of scans into a consistent anatomical coordinate system), which assume image similarity to a reference scan, may fail when applied to these images. However, affine registration is a particularly vital pre-processing step for subsequent image analysis which is assumed to be an effective procedure in recent literature describing sophisticated techniques such as manifold learning. Therefore, in this paper, we present an affine registration solution that uses a graphical model of a population to decompose difficult pairwise registrations into a composition of steps using other members of the population. We developed this methodology in the context of a pre-clinical model of stroke in which large, variable hyper-intense lesions significantly impact registration performance. We tested this technique systematically in a simulated human population of brain tumour images before applying it to pre-clinical models of Parkinson's disease and stroke.

S-HAMMER: hierarchical attribute-guided, symmetric diffeomorphic registration for MR brain images

Deformable registration has been widely used in neuroscience studies for spatial normalization of brain images onto the standard space. Because of possible large anatomical differences across different individual brains, registration performance could be limited when trying to estimate a single directed deformation pathway, i.e., either from template to subject or from subject to template. Symmetric image registration, however, offers an effective way to simultaneously deform template and subject images toward each other until they meet at the middle point. Although some intensity-based registration algorithms have nicely incorporated this concept of symmetric deformation, the pointwise intensity matching between two images may not necessarily imply the matching of correct anatomical correspondences. Based on HAMMER registration algorithm (Shen and Davatzikos, [2002]: IEEE Trans Med Imaging 21:1421-1439), we integrate the strategies of hierarchical attribute matching and symmetric diffeomorphic deformation to build a new symmetric-diffeomorphic HAMMER registration algorithm, called as S-HAMMER. The performance of S-HAMMER has been extensively compared with 14 state-of-the-art nonrigid registration algorithms evaluated in (Klein et al., [2009]: NeuroImage 46:786-802) by using real brain images in LPBA40, IBSR18, CUMC12, and MGH10 datasets. In addition, the registration performance of S-HAMMER, by comparison with other methods, is also demonstrated on both elderly MR brain images (>70 years old) and the simulated brain images with ground-truth deformation fields. In all experiments, our proposed method achieves the best registration performance over all other registration methods, indicating the high applicability of our method in future neuroscience and clinical applications.

Segmentation of ultrasonic breast tumors based on homogeneous patch

PURPOSE

Accurately segmenting breast tumors in ultrasound (US) images is a difficult problem due to their specular nature and appearance of sonographic tumors. The current paper presents a variant of the normalized cut (NCut) algorithm based on homogeneous patches (HP-NCut) for the segmentation of ultrasonic breast tumors.

METHODS

A novel boundary-detection function is defined by combining texture and intensity information to find the fuzzy boundaries in US images. Subsequently, based on the precalculated boundary map, an adaptive neighborhood according to image location referred to as a homogeneous patch (HP) is proposed. HPs are guaranteed to spread within the same tissue region; thus, the statistics of primary features within the HPs is more reliable in distinguishing the different tissues and benefits subsequent segmentation. Finally, the fuzzy distribution of textons within HPs is used as final image features, and the segmentation is obtained using the NCut framework.

RESULTS

The HP-NCut algorithm was evaluated on a large dataset of 100 breast US images (50 benign and 50 malignant). The mean Hausdorff distance measure, the mean minimum Euclidean distance measure and similarity measure achieved 7.1 pixels, 1.58 pixels, and 86.67%, respectively, for benign tumors while those achieved 10.57 pixels, 1.98 pixels, and 84.41%, respectively, for malignant tumors.

CONCLUSIONS

The HP-NCut algorithm provided the improvement in accuracy and robustness compared with state-of-the-art methods. A conclusion that the HP-NCut algorithm is suitable for ultrasonic tumor segmentation problems can be drawn.

Groupwise registration of multimodal images by an efficient joint entropy minimization scheme

Groupwise registration is concerned with bringing a group of images into the best spatial alignment. If images in the group are from different modalities, then the intensity correspondences across the images can be modeled by the joint density function (JDF) of the cooccurring image intensities. We propose a so-called treecode registration method for groupwise alignment of multimodal images that uses a hierarchical intensity-space subdivision scheme through which an efficient yet sufficiently accurate estimation of the (high-dimensional) JDF based on the Parzen kernel method is computed. To simultaneously align a group of images, a gradient-based joint entropy minimization was employed that also uses the same hierarchical intensity-space subdivision scheme. If the Hilbert kernel is used for the JDF estimation, then the treecode method requires no data-dependent bandwidth selection and is thus fully automatic. The treecode method was compared with the ensemble clustering (EC) method on four different publicly available multimodal image data sets and on a synthetic monomodal image data set. The obtained results indicate that the treecode method has similar and, for two data sets, even superior performances compared to the EC method in terms of registration error and success rate. The obtained good registration performances can be mostly attributed to the sufficiently accurate estimation of the JDF, which is computed through the hierarchical intensity-space subdivision scheme, that captures all the important features needed to detect the correct intensity correspondences across a multimodal group of images undergoing registration.

PopTract: population-based tractography

White matter fiber tractography plays a key role in the in vivo understanding of brain circuitry. For tract-based comparison of a population of images, a common approach is to first generate an atlas by averaging, after spatial normalization, all images in the population, and then perform tractography using the constructed atlas. The reconstructed fiber trajectories form a common geometry onto which diffusion properties of each individual subject can be projected based on the corresponding locations in the subject native space. However, in the case of high angular resolution diffusion imaging (HARDI), where modeling fiber crossings is an important goal, the above-mentioned averaging method for generating an atlas results in significant error in the estimation of local fiber orientations and causes a major loss of fiber crossings. These limitatitons have significant impact on the accuracy of the reconstructed fiber trajectories and jeopardize subsequent tract-based analysis. As a remedy, we present in this paper a more effective means of performing tractography at a population level. Our method entails determining a bipolar Watson distribution at each voxel location based on information given by all images in the population, giving us not only the local principal orientations of the fiber pathways, but also confidence levels of how reliable these orientations are across subjects. The distribution field is then fed as an input to a probabilistic tractography framework for reconstructing a set of fiber trajectories that are consistent across all images in the population. We observe that the proposed method, called PopTract, results in significantly better preservation of fiber crossings, and hence yields better trajectory reconstruction in the atlas space.

Feature-based groupwise registration by hierarchical anatomical correspondence detection

Groupwise registration has been widely investigated in recent years due to its importance in analyzing population data in many clinical applications. To our best knowledge, most of the groupwise registration algorithms only utilize the intensity information. However, it is well known that using intensity only is not sufficient to achieve the anatomically sound correspondences in medical image registration. In this article, we propose a novel feature-based groupwise registration algorithm to establish the anatomical correspondence across subjects by using the attribute vector that is defined as the morphological signature for each voxel. Similar to most of the state-of-the-art groupwise registration algorithms, which simultaneously estimate the transformation fields for all subjects, we develop an energy function to minimize the intersubject discrepancies on anatomical structures and drive all subjects toward the hidden common space. To make the algorithm efficient and robust, we decouple the complex groupwise registration problem into two easy-to-solve subproblems, namely (1) robust correspondence detection and (2) dense transformation field estimation, which are systematically integrated into a unified framework. To achieve the robust correspondences in the step (1), several strategies are adopted. First, the procedure of feature matching is evaluated within a neighborhood, rather than only on a single voxel. Second, the driving voxels with distinctive image features are designed to drive the transformations of other nondriving voxels. Third, we take advantage of soft correspondence assignment not only in the spatial domain but also across the population of subjects. Specifically, multiple correspondences are allowed to alleviate the ambiguity in establishing correspondences w.r.t. a particular subject and also the contributions from different subjects are dynamically controlled throughout the registration. Eventually in the step (2), based on the correspondences established for the driving voxels, thin-plate spline is used to propagate correspondences on the driving voxels to other locations in the image. By iteratively repeating correspondence detection and dense deformation estimation, all the subjects will be aligned onto the common space. Our feature-based groupwise registration algorithm has been extensively evaluated over 18 elderly brains, 16 brains from NIREP (with 32 manually delineated labels), 40 brains from LONI LPBA40 (with 54 manually delineated labels), and 12 pairs of normal controls and simulated atrophic brain images. In all experiments, our algorithm achieves more robust and accurate registration results, compared with another groupwise algorithm and a pairwise registration method.