Skip to main content
Fachgebiete
Automobil + Motoren Bauwesen + Immobilien Business IT + Informatik Elektrotechnik + Elektronik Energie + Nachhaltigkeit Finance + Banking Management + Führung Marketing + Vertrieb Maschinenbau + Werkstoffe Versicherung + Risiko
DE
EN
Themenseiten
Organisationspsychologie Projektmanagement Marketing Smart Manufacturing
Bücher
Zeitschriften
Weitere Formate
Podcasts Webinare Technik Webinare Wirtschaft Kongresse Veranstaltungskalender Awards
Jetzt Einzelzugang starten
Zugang für Unternehmen
Referenzkunden
MTZ-Fachkongress

Antriebe und Energiesysteme von morgen


Austausch von Automobil- und Energiewirtschaft

Am 14./15. Mai geht es in Chemnitz um die Mobilitätswende durch Elektro- und Wasserstofffahrzeuge.
Erweiterte Suche
Anmelden
nach oben

2017 | Buch

Constructing Subject- and Disease-Specific Effect Maps: Application to Neurodegenerative Diseases Kapitel lesen Erstes Kapitel lesen

Medical Computer Vision and Bayesian and Graphical Models for Biomedical Imaging

MICCAI 2016 International Workshops, MCV and BAMBI, Athens, Greece, October 21, 2016, Revised Selected Papers

herausgegeben von: Henning Müller, B. Michael Kelm, Tal Arbel, Weidong Cai, M. Jorge Cardoso, Georg Langs, Bjoern Menze, Dimitris Metaxas, Albert Montillo, William M. Wells III, Shaoting Zhang, Albert C.S. Chung, Mark Jenkinson, Annemie Ribbens

Verlag: Springer International Publishing

Buchreihe : Lecture Notes in Computer Science

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten
insite
SUCHEN

Über dieses Buch

This book constitutes the thoroughly refereed post-workshop proceedings of the International Workshop on Medical Computer Vision, MCV 2016, and of the International Workshop on Bayesian and grAphical Models for Biomedical Imaging, BAMBI 2016, held in Athens, Greece, in October 2016, held in conjunction with the 19th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2016.
The 13 papers presented in MCV workshop and the 6 papers presented in BAMBI workshop were carefully reviewed and selected from numerous submissions.
The goal of the MCV workshop is to explore the use of "big data” algorithms for harvesting, organizing and learning from large-scale medical imaging data sets and for general-purpose automatic understanding of medical images.
The BAMBI workshop aims to highlight the potential of using Bayesian or random field graphical models for advancing research in biomedical image analysis.

Inhaltsverzeichnis

Frontmatter

MCV Workshop: Brain Imaging

Frontmatter
Constructing Subject- and Disease-Specific Effect Maps: Application to Neurodegenerative Diseases
Abstract
Current statistical methods in neuroimaging identify effects of neurodegenerative diseases on the brain structure by detecting group differences. Results are detailed maps showing population-wide effects. Although useful for better understanding the disease, these maps provide little subject-specific information. Furthermore, since group assignments have to be known prior to analysis, resulting maps have limited diagnostic value for new subjects. This article proposes a method to construct subject- and disease-specific effect maps prior to diagnosis. The method combines techniques from binary classification and image restoration to identify the effects of a disease of interest on the measurements. Experimental evaluation is carried out with synthetically generated data and real data selected from the ADNI cohort. Results demonstrate the capability of the proposed method in generating subject-specific effect maps.
Ender Konukoglu, Ben Glocker
BigBrain: Automated Cortical Parcellation and Comparison with Existing Brain Atlases
Abstract
Most available 3D human brain atlases provide information only at a macroscopic level, while 2D atlases are often at a microscopic level but lack 3D integration. A 3D atlas defined upon fine-grain anatomical detail of cortical layers and cells is necessary to fully understand neurobiological processes. “BigBrain,” a high-resolution 3D model of a human brain at nearly cellular resolution, was released in 2013. This unique dataset enables the extraction of microscopic data for utilization in brain mapping, modeling and simulation. We propose an automated 3D cortical parcellation of the BigBrain volume into functionally-meaningful areas in order to create a modern high-resolution 3D cytoarchitectural atlas that will complement existing brain atlases. We use a distance metrics-based framework for BigBrain parcellation, and perform comparative analyses of our results with existing atlases (Brodmann and JuBrain atlases). This work has immediate application in teaching, neurosurgery, cognitive neuroscience, and imaging-based brain mapping.
Marc Fournier, Lindsay B. Lewis, Alan C. Evans
LATEST: Local AdapTivE and Sequential Training for Tissue Segmentation of Isointense Infant Brain MR Images
Abstract
Accurate segmentation of isointense infant (~6 months of age) brain MRIs is of great importance, however, a very challenging task, due to extremely low tissue contrast caused by ongoing myelination processes. In this work, we propose a novel learning method based on Local AdapTivE and Sequential Training (LATEST) for segmentation. Specifically, random forest technique is employed to train a local classifier (a single decision tree) for each voxel in the common space based on the neighboring training samples from atlases. Then, for each given voxel, all trained nearby individual classifiers (decision trees) are grouped together to form a forest. Moreover, the estimated probabilities are further used as additional source images to train the next set of local classifiers for refining tissue classification. By iteratively training the subsequent classifiers based on the updated tissue probability maps, a sequence of local classifiers can be built for accurate tissue segmentation.
Li Wang, Yaozong Gao, Gang Li, Feng Shi, Weili Lin, Dinggang Shen
Landmark-Based Alzheimer’s Disease Diagnosis Using Longitudinal Structural MR Images
Abstract
In this paper, we propose a landmark-based feature extraction method for AD diagnosis using longitudinal structural MR images, which requires no nonlinear registration or tissue segmentation in the application stage and is robust to the inconsistency among longitudinal scans. Specifically, (1) the discriminative landmarks are first automatically discovered from the whole brain, which can be efficiently localized using a fast landmark detection method for the testing images; (2) High-level statistical spatial features and contextual longitudinal features are then extracted based on those detected landmarks. Using the spatial and longitudinal features, a linear support vector machine (SVM) is adopted for distinguishing AD subjects from healthy controls (HCs) and also mild cognitive impairment (MCI) subjects from HCs, respectively. Experimental results demonstrate the competitive classification accuracies, as well as a promising computational efficiency.
Jun Zhang, Mingxia Liu, Le An, Yaozong Gao, Dinggang Shen

MCV Workshop: Lung Imaging

Frontmatter
Inferring Disease Status by Non-parametric Probabilistic Embedding
Abstract
Computing similarity between all pairs of patients in a dataset enables us to group the subjects into disease subtypes and infer their disease status. However, robust and efficient computation of pairwise similarity is a challenging task for large-scale medical image datasets. We specifically target diseases where multiple subtypes of pathology present simultaneously, rendering the definition of the similarity a difficult task. To define pairwise patient similarity, we characterize each subject by a probability distribution that generates its local image descriptors. We adopt a notion of affinity between probability distributions which lends itself to similarity between subjects. Instead of approximating the distributions by a parametric family, we propose to compute the affinity measure indirectly using an approximate nearest neighbor estimator. Computing pairwise similarities enables us to embed the entire patient population into a lower dimensional manifold, mapping each subject from high-dimensional image space to an informative low-dimensional representation. We validate our method on a large-scale lung CT scan study and demonstrate the state-of-the-art prediction on an important physiologic measure of airflow (the forced expiratory volume in one second, FEV1) in addition to a 5-category clinical rating (so-called GOLD score).
Nematollah Kayhan Batmanghelich, Ardavan Saeedi, Raul San Jose Estepar, Michael Cho, William M. Wells III
A Lung Graph–Model for Pulmonary Hypertension and Pulmonary Embolism Detection on DECT Images
Abstract
This article presents a novel graph–model approach encoding the relations between the perfusion in several regions of the lung extracted from a geometry–based atlas. Unlike previous approaches that individually analyze regions of the lungs, our method evaluates the entire pulmonary circulatory network for the classification of patients with pulmonary embolism and pulmonary hypertension. An undirected weighted graph with fixed structure is used to encode the network of intensity distributions in Dual Energy Computed Tomography (DECT) images. Results show that the graph–model presented is capable of characterizing a DECT dataset of 30 patients affected with disease and 26 healthy patients, achieving a discrimination accuracy from 0.77 to 0.87 and an AUC between 0.73 and 0.86. This fully automatic graph–model of the lungs constitutes a novel and effective approach for exploring the various patterns of pulmonary perfusion of healthy and diseased patients.
Yashin Dicente Cid, Henning Müller, Alexandra Platon, Jean–Paul Janssens, Frédéric Lador, Pierre–Alexandre Poletti, Adrien Depeursinge
Explaining Radiological Emphysema Subtypes with Unsupervised Texture Prototypes: MESA COPD Study
Abstract
Pulmonary emphysema is traditionally subcategorized into three subtypes, which have distinct radiological appearances on computed tomography (CT) and can help with the diagnosis of chronic obstructive pulmonary disease (COPD). Automated texture-based quantification of emphysema subtypes has been successfully implemented via supervised learning of these three emphysema subtypes. In this work, we demonstrate that unsupervised learning on a large heterogeneous database of CT scans can generate texture prototypes that are visually homogeneous and distinct, reproducible across subjects, and capable of predicting accurately the three standard radiological subtypes. These texture prototypes enable automated labeling of lung volumes, and open the way to new interpretations of lung CT scans with finer subtyping of emphysema.
Jie Yang, Elsa D. Angelini, Benjamin M. Smith, John H. M. Austin, Eric A. Hoffman, David A. Bluemke, R. Graham Barr, Andrew F. Laine

MCV Workshop: Segmentation, Detection, and Classification

Frontmatter
Automatic Segmentation of Abdominal MRI Using Selective Sampling and Random Walker
Abstract
MRI segmentation is a challenging task due to low anatomical contrast and large inter-patient variation. We propose a feature-driven automatic segmentation framework, combining voxel-wise classification with a Random-Walker (RW) based spatial regularization. Typically, such steps are treated independently, i.e. classification outcome is maximized without taking into account the regularization to follow. Herein we present a method for selective sampling of training patches, in view of the posterior spatial regularization. This aims to concentrate training samples near desired anatomical boundaries, around which the gain from a subsequent RW regularization will potentially be minimal. This trades off a lower classification accuracy for a higher joint segmentation performance. We compare our proposed sampling strategy to conventional uniform sampling on 20 full-body MR T1 scans from the VISCERAL dataset, both with RW and Markov Random Fields regularizations, showing Dice improvements of up to 12\(\times \) with the proposed approach.
Janine Thoma, Firat Ozdemir, Orcun Goksel
Gaze2Segment: A Pilot Study for Integrating Eye-Tracking Technology into Medical Image Segmentation
Abstract
In this study, we developed a novel system, called Gaze2Segment, integrating biological and computer vision techniques to support radiologists’ reading experience with an automatic image segmentation task. During diagnostic assessment of lung CT scans, the radiologists’ gaze information were used to create a visual attention map. Next, this map was combined with a computer-derived saliency map, extracted from the gray-scale CT images. The visual attention map was used as an input for indicating roughly the location of a region of interest. With computer-derived saliency information, on the other hand, we aimed at finding foreground and background cues for the object of interest found in the previous step. These cues are used to initiate a seed-based delineation process. The proposed Gaze2Segment achieved a dice similarity coefficient of 86% and Hausdorff distance of 1.45 mm as a segmentation accuracy. To the best of our knowledge, Gaze2Segment is the first true integration of eye-tracking technology into a medical image segmentation task without the need for any further user-interaction.
Naji Khosravan, Haydar Celik, Baris Turkbey, Ruida Cheng, Evan McCreedy, Matthew McAuliffe, Sandra Bednarova, Elizabeth Jones, Xinjian Chen, Peter Choyke, Bradford Wood, Ulas Bagci
Automatic Detection of Histological Artifacts in Mouse Brain Slice Images
Abstract
A major challenge in automatic registration, alignment and 3-D reconstruction of conventionally processed mouse brain slice images is the presence of histological artifacts, like tissue tears and losses. These artifacts are often produced from manual sample preparation processes, which are ubiquitous in most neuroanatomical laboratories. We present a novel geometric algorithm to automatically detect these artifacts (damage regions) in mouse brain slice images. Our algorithm is guided by our observation that the tears and tissue loss in brain slice images result in external geometric medial axis of the outer contours to go deep inside the tissue. We tested our algorithm on 52 mouse brain slice images with major histological artifacts and successfully detected all the damage regions in the dataset. Our algorithm also demonstrated much lower errors when quantitatively evaluated by performing feature based registration between all 52 slices and their corresponding Allen Reference Atlas (ARA) images.
Nitin Agarwal, Xiangmin Xu, M. Gopi
Lung Nodule Classification by Jointly Using Visual Descriptors and Deep Features
Abstract
Classifying benign and malignant lung nodules using the thoracic computed tomography (CT) screening is the primary method for early diagnosis of lung cancer. Despite of their widely recognized success in image classification, deep learning techniques may not achieve satisfying accuracy on this problem, due to the limited training samples resulted from the all-consuming nature of medical image acquisition and annotation. In this paper, we jointly use the texture and shape descriptors, which characterize the heterogeneity of nodules, and the features learned by a deep convolutional neural network, and thus proposed a combined-feature based classification (CFBC) algorithm to differentiate lung nodules. We have evaluated this algorithm against four state-of-the-art nodule classification approaches on the benchmark LIDC-IDRI dataset. Our results suggest that the proposed CFBC algorithm can distinguish malignant lung nodules from benign ones more accurately than other four methods.
Yutong Xie, Jianpeng Zhang, Sidong Liu, Weidong Cai, Yong Xia
Representation Learning for Cross-Modality Classification
Abstract
Differences in scanning parameters or modalities can complicate image analysis based on supervised classification. This paper presents two representation learning approaches, based on autoencoders, that address this problem by learning representations that are similar across domains. Both approaches use, next to the data representation objective, a similarity objective to minimise the difference between representations of corresponding patches from each domain. We evaluated the methods in transfer learning experiments on multi-modal brain MRI data and on synthetic data. After transforming training and test data from different modalities to the common representations learned by our methods, we trained classifiers for each of pair of modalities. We found that adding the similarity term to the standard objective can produce representations that are more similar and can give a higher accuracy in these cross-modality classification experiments.
Gijs van Tulder, Marleen de Bruijne
Guideline-Based Machine Learning for Standard Plane Extraction in 3D Cardiac Ultrasound
Abstract
The extraction of six standard planes in 3D cardiac ultrasound plays an important role in clinical examination to analyze cardiac function. This paper proposes a guideline-based machine learning method for efficient and accurate standard plane extraction. A cardiac ultrasound guideline determines appropriate operation steps for clinical examinations. The idea of guideline-based machine learning is incorporating machine learning approaches into each stage of the guideline. First, Hough forest with hierarchical search is applied for 3D feature point detection. Second, initial planes are determined using anatomical regularities according to the guideline. Finally, a regression forest integrated with constraints of plane regularities is applied for refining each plane. The proposed method was evaluated on a 3D cardiac ultrasound dataset. Compared with other plane extraction methods, it demonstrated an improved accuracy with a significantly faster running time of 0.8 s per volume.
Peifei Zhu, Zisheng Li

BAMBI Workshop

Frontmatter
A Statistical Model for Simultaneous Template Estimation, Bias Correction, and Registration of 3D Brain Images
Abstract
Template estimation plays a crucial role in computational anatomy since it provides reference frames for performing statistical analysis of the underlying anatomical population variability. While building models for template estimation, variability in sites and image acquisition protocols need to be accounted for. To account for such variability, we propose a generative template estimation model that makes simultaneous inference of both bias fields in individual images, deformations for image registration, and variance hyperparameters. In contrast, existing maximum a posterori based methods need to rely on either bias-invariant similarity measures or robust image normalization. Results on synthetic and real brain MRI images demonstrate the capability of the model to capture heterogeneity in intensities and provide a reliable template estimation from registration.
Akshay Pai, Stefan Sommer, Lars Lau Raket, Line Kühnel, Sune Darkner, Lauge Sørensen, Mads Nielsen
Bayesian Multiview Manifold Learning Applied to Hippocampus Shape and Clinical Score Data
Abstract
In this paper, we present a novel Bayesian model for manifold learning, suitable for data that are comprised of multiple modes of observations. Our data are assumed to be lying on a non-linear, low-dimensional manifold, modelled as a locally linear structure. The manifold local structure and the manifold coordinates are latent stochastic variables that are estimated from a training set. Through the use of appropriate prior distributions, neighbouring points are constrained to have similar manifold coordinates as well as similar manifold geometry. A single set of latent coordinates is learned, common for all views. We show how to solve the model with variational inference. We also exploit the multiview aspect of the proposed model, by showing how to estimate missing views of unseen data. We have tested the proposed model and methods on medical imaging data of the OASIS brain MRI dataset [6]. The data are comprised of four views: two views that correspond to clinical scores and two views that correspond to hippocampus shape extracted from the OASIS MR images. Our model is successfully used to map the multimodal data to probabilistic embedding coordinates, as well as estimate missing clinical scores and shape information of test data.
Giorgos Sfikas, Christophoros Nikou
Rigid Slice-To-Volume Medical Image Registration Through Markov Random Fields
Abstract
Rigid slice-to-volume registration is a challenging task, which finds application in medical imaging problems like image fusion for image guided surgeries and motion correction for volume reconstruction. It is usually formulated as an optimization problem and solved using standard continuous methods. In this paper, we discuss how this task be formulated as a discrete labeling problem on a graph. Inspired by previous works on discrete estimation of linear transformations using Markov Random Fields (MRFs), we model it using a pairwise MRF, where the nodes are associated to the rigid parameters, and the edges encode the relation between the variables. We compare the performance of the proposed method to a continuous formulation optimized using simplex, and we discuss how it can be used to further improve the accuracy of our approach. Promising results are obtained using a monomodal dataset composed of magnetic resonance images (MRI) of a beating heart.
Roque Porchetto, Franco Stramana, Nikos Paragios, Enzo Ferrante
Sparse Probabilistic Parallel Factor Analysis for the Modeling of PET and Task-fMRI Data
Abstract
Modern datasets are often multiway in nature and can contain patterns common to a mode of the data (e.g. space, time, and subjects). Multiway decomposition such as parallel factor analysis (PARAFAC) take into account the intrinsic structure of the data, and sparse versions of these methods improve interpretability of the results. Here we propose a variational Bayesian parallel factor analysis (VB-PARAFAC) model and an extension with sparse priors (SP-PARAFAC). Notably, our formulation admits time and subject specific noise modeling as well as subject specific offsets (i.e., mean values). We confirmed the validity of the models through simulation and performed exploratory analysis of positron emission tomography (PET) and functional magnetic resonance imaging (fMRI) data. Although more constrained, the proposed models performed similarly to more flexible models in approximating the PET data, which supports its robustness against noise. For fMRI, both models correctly identified task-related components, but were not able to segregate overlapping activations.
Vincent Beliveau, Georgios Papoutsakis, Jesper Løve Hinrich, Morten Mørup
Non-local Graph-Based Regularization for Deformable Image Registration
Abstract
Deformable image registration aims to deliver a plausible spatial transformation between two or more images by solving a highly dimensional, ill-posed optimization problem. Covering the complexity of physiological motion has so far been limited to either generic physical models or local motion regularization models. This paper presents an alternative, graphical regularization model, which captures well the non-local scale of motion, and thus enables to incorporate complex regularization models directly into deformable image registration. In order to build the proposed graph-based regularization, a Minimum Spanning Tree (MST), which represents the underlying tissue physiology in a perceptually meaningful way, is computed first. This is followed by a fast non-local cost aggregation algorithm that performs regularization of the estimated displacement field using the precomputed MST. To demonstrate the advantage of the presented regularization, we embed it into the widely used Demons registration framework. The presented method is shown to improve the accuracy for exhale-inhale CT data pairs.
Bartłomiej W. Papież, Adam Szmul, Vicente Grau, J. Michael Brady, Julia A. Schnabel
Unsupervised Framework for Consistent Longitudinal MS Lesion Segmentation
Abstract
Quantification of white matter lesion changes on brain magnetic resonance (MR) images is of major importance for the follow-up of patients with Multiple Sclerosis (MS). Many automated segmentation methods have been proposed. However, most of them focus on a single time point MR scan session and hence lack consistency when evaluating lesion changes over time. In this paper, we present MSmetrix-long, an unsupervised method that incorporates temporal consistency by jointly segmenting MS lesions of two subsequent scan sessions. The method is formulated as a Maximum A Posteriori model on the FLAIR image intensities of both time points and the difference image intensities, and optimised using an expectation maximisation algorithm. Validation is performed on two different data sets in terms of consistency and sensitivity to MS lesion changes. It is shown that MSmetrix-long outperforms MSmetrix-cross for the quantification of MS lesion evolution over time.
Saurabh Jain, Annemie Ribbens, Diana M. Sima, Sabine Van Huffel, Frederik Maes, Dirk Smeets
Backmatter
Metadaten
Titel
Medical Computer Vision and Bayesian and Graphical Models for Biomedical Imaging
herausgegeben von
Henning Müller
B. Michael Kelm
Tal Arbel
Weidong Cai
M. Jorge Cardoso
Georg Langs
Bjoern Menze
Dimitris Metaxas
Albert Montillo
William M. Wells III
Shaoting Zhang
Albert C.S. Chung
Mark Jenkinson
Annemie Ribbens
Copyright-Jahr
2017
Electronic ISBN
978-3-319-61188-4
Print ISBN
978-3-319-61187-7
DOI
https://doi.org/10.1007/978-3-319-61188-4

Premium Partner

    Bildnachweise