Using manifold learning for atlas selection in multi-atlas segmentation.
ABSTRACT: Multi-atlas segmentation has been widely used to segment various anatomical structures. The success of this technique partly relies on the selection of atlases that are best mapped to a new target image after registration. Recently, manifold learning has been proposed as a method for atlas selection. Each manifold learning technique seeks to optimize a unique objective function. Therefore, different techniques produce different embeddings even when applied to the same data set. Previous studies used a single technique in their method and gave no reason for the choice of the manifold learning technique employed nor the theoretical grounds for the choice of the manifold parameters. In this study, we compare side-by-side the results given by 3 manifold learning techniques (Isomap, Laplacian Eigenmaps and Locally Linear Embedding) on the same data set. We assess the ability of those 3 different techniques to select the best atlases to combine in the framework of multi-atlas segmentation. First, a leave-one-out experiment is used to optimize our method on a set of 110 manually segmented atlases of hippocampi and find the manifold learning technique and associated manifold parameters that give the best segmentation accuracy. Then, the optimal parameters are used to automatically segment 30 subjects from the Alzheimer's Disease Neuroimaging Initiative (ADNI). For our dataset, the selection of atlases with Locally Linear Embedding gives the best results. Our findings show that selection of atlases with manifold learning leads to segmentation accuracy close to or significantly higher than the state-of-the-art method and that accuracy can be increased by fine tuning the manifold learning process.
Project description:Multi-atlas brain segmentation of human brain MR images allows quantification research in structural neuroimaging. To achieve high accuracy and computational efficiency of segmentation relies on a custom subset of atlases for each target subject. However, the criterion for atlas pre-selection remains an open question. In this study, two atlas pre-selection approaches based on location-based feature matching were proposed and compared to random and mutual information-based methods using a database of 47 atlases. A varying number of atlases ranked top with hierarchical structural granularity were compared using Dice overlap. The results indicated that the proposed 4L approach consistently led to the highest level of accuracy at a given number of employed atlases in both adult and geriatric populations. In addition, the proposed two methods (4L and LV) can reduce 20 times computational time compared with the stereotypical mutual information-based method. Our pre-selection strategy would provide better segmentation performance in terms of both accuracy and efficiency. The proposed atlas pre-selection will be further implemented into our online automatic brain image segmentation system (www.mricloud.org).
Project description:Previous work from our group demonstrated the use of multiple input atlases to a modified multi-atlas framework (MAGeT-Brain) to improve subject-based segmentation accuracy. Currently, segmentation of the striatum, globus pallidus and thalamus are generated from a single high-resolution and -contrast MRI atlas derived from annotated serial histological sections. Here, we warp this atlas to five high-resolution MRI templates to create five de novo atlases. The overall goal of this work is to use these newly warped atlases as input to MAGeT-Brain in an effort to consolidate and improve the workflow presented in previous manuscripts from our group, allowing for simultaneous multi-structure segmentation. The work presented details the methodology used for the creation of the atlases using a technique previously proposed, where atlas labels are modified to mimic the intensity and contrast profile of MRI to facilitate atlas-to-template nonlinear transformation estimation. Dice's Kappa metric was used to demonstrate high quality registration and segmentation accuracy of the atlases. The final atlases are available at https://github.com/CobraLab/atlases/tree/master/5-atlas-subcortical.
Project description:BACKGROUND AND PURPOSE:The aim of this study was to evaluate an automatic multi-atlas-based segmentation method for generating prostate, peripheral (PZ), and transition zone (TZ) contours on MRIs with and without fat saturation (±FS), and compare MRIs from different vendor MRI systems. METHODS:T2-weighted (T2) and fat-saturated (T2FS) MRIs were acquired on 3T GE (GE, Waukesha, WI, USA) and Siemens (Erlangen, Germany) systems. Manual prostate and PZ contours were used to create atlas libraries. As a test MRI is entered, the procedure for atlas segmentation automatically identifies the atlas subjects that best match the test subject, followed by a normalized intensity-based free-form deformable registration. The contours are transformed to the test subject, and Dice similarity coefficients (DSC) and Hausdorff distances between atlas-generated and manual contours were used to assess performance. RESULTS:Three atlases were generated based on GE_T2 (n?=?30), GE_T2FS (n?=?30), and Siem_T2FS (n?=?31). When test images matched the contrast and vendor of the atlas, DSCs of 0.81 and 0.83 for T2?±?FS were obtained (baseline performance). Atlases performed with higher accuracy when segmenting (i) T2FS vs. T2 images, likely due to a superior contrast between prostate vs. surrounding tissue; (ii) prostate vs. zonal anatomy; (iii) in the mid-gland vs. base and apex. Atlases performance declined when tested with images with differing contrast and MRI vendor. Conversely, combined atlases showed similar performance to baseline. CONCLUSION:The MRI atlas-based segmentation method achieved good results for prostate, PZ, and TZ compared to expert contoured volumes. Combined atlases performed similarly to matching atlas and scan type. The technique is fast, fully automatic, and implemented on commercially available clinical platform.
Project description:Technologies for multi-atlas brain segmentation of T1-weighted MRI images have rapidly progressed in recent years, with highly promising results. This approach, however, relies on a large number of atlases with accurate and consistent structural identifications. Here, we introduce our atlas inventories (n=90), which cover ages 4-82years with unique hierarchical structural definitions (286 structures at the finest level). This multi-atlas library resource provides the flexibility to choose appropriate atlases for various studies with different age ranges and structure-definition criteria. In this paper, we describe the details of the atlas resources and demonstrate the improved accuracy achievable with a dynamic age-matching approach, in which atlases that most closely match the subject's age are dynamically selected. The advanced atlas creation strategy, together with atlas pre-selection principles, is expected to support the further development of multi-atlas image segmentation.
Project description:BACKGROUND:Multi-atlas segmentation, a popular technique implemented in the Automated Segmentation of Hippocampal Subfields (ASHS) software, utilizes multiple expert-labelled images ("atlases") to delineate medial temporal lobe substructures. This multi-atlas method is increasingly being employed in early Alzheimer's disease (AD) research, it is therefore becoming important to know how the construction of the atlas set in terms of proportions of controls and patients with mild cognitive impairment (MCI) and/or AD affects segmentation accuracy. OBJECTIVE:To evaluate whether the proportion of controls in the training sets affects the segmentation accuracy of both controls and patients with MCI and/or early AD at 3T and 7T. METHODS:We performed cross-validation experiments varying the proportion of control subjects in the training set, ranging from a patient-only to a control-only set. Segmentation accuracy of the test set was evaluated by the Dice similarity coeffiecient (DSC). A two-stage statistical analysis was applied to determine whether atlas composition is linked to segmentation accuracy in control subjects and patients, for 3T and 7T. RESULTS:The different atlas compositions did not significantly affect segmentation accuracy at 3T and for patients at 7T. For controls at 7T, including more control subjects in the training set significantly improves the segmentation accuracy, but only marginally, with the maximum of 0.0003 DSC improvement per percent increment of control subject in the training set. CONCLUSION:ASHS is robust in this study, and the results indicate that future studies investigating hippocampal subfields in early AD populations can be flexible in the selection of their atlas compositions.
Project description:In this paper, we present a multi-atlas-based framework for accurate, consistent and simultaneous segmentation of a group of target images. Multi-atlas-based segmentation algorithms consider concurrently complementary information from multiple atlases to produce optimal segmentation outcomes. However, the accuracy of these algorithms relies heavily on the precise alignment of the atlases with the target image. In particular, the commonly used pairwise registration may result in inaccurate alignment especially between images with large shape differences. Additionally, when segmenting a group of target images, most current methods consider these images independently with disregard of their correlation, thus resulting in inconsistent segmentations of the same structures across different target images. We propose two novel strategies to address these limitations: 1) a novel tree-based groupwise registration method for concurrent alignment of both the atlases and the target images, and 2) an iterative groupwise segmentation method for simultaneous consideration of segmentation information propagated from all available images, including the atlases and other newly segmented target images. Evaluation based on various datasets indicates that the proposed multi-atlas-based multi-image segmentation (MABMIS) framework yields substantial improvements in terms of consistency and accuracy over methods that do not consider the group of target images holistically.
Project description:Using multi-atlas registration (MAR), information carried by atlases can be transferred onto a new input image for the tasks of region of interest (ROI) segmentation, anatomical landmark detection, and so on. Conventional atlases used in MAR methods are monomodal and contain only normal anatomical structures. Therefore, the majority of MAR methods cannot handle input multimodal pathological images, which are often collected in routine image-based diagnosis. This is because registering monomodal atlases with normal appearances to multimodal pathological images involves two major problems: (1) missing imaging modalities in the monomodal atlases, and (2) influence from pathological regions. In this paper, we propose a new MAR framework to tackle these problems. In this framework, a deep learning based image synthesizers are applied for synthesizing multimodal normal atlases from conventional monomodal normal atlases. To reduce the influence from pathological regions, we further propose a multimodal lowrank approach to recover multimodal normal-looking images from multimodal pathological images. Finally, the multimodal normal atlases can be registered to the recovered multimodal images in a multi-channel way. We evaluate our MAR framework via brain ROI segmentation of multimodal tumor brain images. Due to the utilization of multimodal information and the reduced influence from pathological regions, experimental results show that registration based on our method is more accurate and robust, leading to significantly improved brain ROI segmentation compared with state-of-the-art methods.
Project description:Atlas-based automated anatomical labeling is a fundamental tool in medical image segmentation, as it defines regions of interest for subsequent analysis of structural and functional image data. The extensive investigation of multi-atlas warping and fusion techniques over the past 5 or more years has clearly demonstrated the advantages of consensus-based segmentation. However, the common approach is to use multiple atlases with a single registration method and parameter set, which is not necessarily optimal for every individual scan, anatomical region, and problem/data-type. Different registration criteria and parameter sets yield different solutions, each providing complementary information. Herein, we present a consensus labeling framework that generates a broad ensemble of labeled atlases in target image space via the use of several warping algorithms, regularization parameters, and atlases. The label fusion integrates two complementary sources of information: a local similarity ranking to select locally optimal atlases and a boundary modulation term to refine the segmentation consistently with the target image's intensity profile. The ensemble approach consistently outperforms segmentations using individual warping methods alone, achieving high accuracy on several benchmark datasets. The MUSE methodology has been used for processing thousands of scans from various datasets, producing robust and consistent results. MUSE is publicly available both as a downloadable software package, and as an application that can be run on the CBICA Image Processing Portal (https://ipp.cbica.upenn.edu), a web based platform for remote processing of medical images.
Project description:Three-dimensional digital atlases are rapidly becoming indispensible in modern biology. We used serial sectioning combined with manual registration and segmentation of images to develop a comprehensive and detailed three-dimensional atlas of the honeybee head-neck system. This interactive atlas includes skeletal structures of the head and prothorax, the neck musculature, and the nervous system. The scope and resolution of the model exceeds atlases previously developed on similar sized animals, and the interactive nature of the model provides a far more accessible means of interpreting and comprehending insect anatomy and neuroanatomy.
Project description:The thalamic nuclei are involved in many neurodegenerative diseases and therefore, their identification is of key importance in numerous clinical treatments. Automated segmentation of thalamic subparts is currently achieved by exploring diffusion-weighted magnetic resonance imaging (DW-MRI), but in absence of such data, atlas-based segmentation can be used as an alternative. Currently, there is a limited number of available digital atlases of the thalamus. Moreover, all atlases are created using a few subjects only, thus are prone to errors due to the inter-subject variability of the thalamic morphology. In this work, we present a probabilistic atlas of anatomical subparts of the thalamus built upon a relatively large dataset where the individual thalamic parcellation was done by employing a recently proposed automatic diffusion-based clustering method. Our analyses, comparing the segmentation performance between the atlas-based and the clustering method, demonstrate the ability of the provided atlas to substitute the automated diffusion-based subdivision in the individual space when the DW-MRI is not available.