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
Bücher
Zeitschriften
Themenseiten
Organisationspsychologie Projektmanagement Marketing Smart Manufacturing
Weitere Formate
Podcasts Webinare Technik Webinare Wirtschaft Kongresse Veranstaltungskalender Awards
Jetzt Einzelzugang starten
Zugang für Unternehmen
Referenzkunden
SLX-Digitalkonferenz: Zukunftswerkstatt 2024

Mehr Umsatz mit Top-Kontakten

Am 7. Mai erfahren Sie, wie Vertriebsteams Leadgenerierung, Leadnurturing und Leadstrategien effizient steuern können.
Erweiterte Suche
Anmelden
nach oben

2015 | Buch

Object Segmentation and Markov Random Fields Kapitel lesen Erstes Kapitel lesen

Handbook of Biomedical Imaging

Methodologies and Clinical Research

herausgegeben von: Nikos Paragios, James Duncan, Nicholas Ayache

Verlag: Springer US

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

Einloggen, um Zugang zu erhalten
insite
SUCHEN

Über dieses Buch

This book explains the process of computer assisted biomedical image analysis diagnosis through mathematical modeling and inference of image-based bio-markers. It covers five crucial thematic areas: methodologies, statistical and physiological models, biomedical perception, clinical biomarkers, and emerging modalities and domains.

The dominant state-of-the-art methodologies for content extraction and interpretation of medical images include fuzzy methods, level set methods, kernel methods, and geometric deformable models. The models and techniques discussed are used in the diagnosis, planning, control and follow-up of medical procedures. Throughout the book, challenges and limitations are explored along with new research directions.

Handbook of Biomedical Imaging offers a unique guide to the entire chain of biomedical imaging, explaining how image formation is done and how the most appropriate techniques are used to address demands and diagnoses. Radiologists, research scientists, senior undergraduate and graduate students in health sciences and engineering, and university professors will find this to be an exceptional reference.

Inhaltsverzeichnis

Frontmatter

Methodologies

Frontmatter
Object Segmentation and Markov Random Fields
Abstract
This chapter discusses relationships between graph cut approach to object delineation and other standard techniques optimizing segmentation boundaries. Graph cut method is presented in the context of globally optimal labeling of binary Markov Random Fields (MRFs). We review algorithms details and show several 2D and 3D examples.
Y Boykov
Fuzzy methods in medical imaging
Abstract
Fuzzy sets theory is of great interest in medical image processing, for dealing with imprecise information and knowledge. It provides a consistent mathematical framework for knowledge representation, information modeling at different levels, fusion of heterogeneous information, reasoning and decision making. In this chapter, we provide an overview of the potential of this theory in medical imaging, in particular for classification, segmentation and recognition of anatomical and pathological structures.
I. Bloch
Curve Propagation, Level Set Methods and Grouping
Abstract
Image segmentation and object extraction are among the most well addressed topics in computational vision. In this chapter we present a comprehensive tutorial of level sets towards a flexible frame partition paradigm that could integrate edge-drive, regional-based and prior knowledge to object extraction. The central idea behind such an approach is to perform image partition through the propagation planar curves/surfaces. To this end, an objective function that aims to account for the expected visual properties of the object, impose certain smoothness constraints and encode prior knowledge on the geometric form of the object to be recovered is presented. Promising experimental results demonstrate the potential of such a method.
N. Paragios
Kernel Methods in Medical Imaging
Abstract
We introduce machine learning techniques, more specifically kernel methods, and show how they can be used for medical imaging. After a tutorial presentation of machine learning concepts and tools, including Support Vector Machine (SVM), kernel ridge regression and kernel PCA, we present an application of these tools to the prediction of Computed Tomography (CT) images based on Magnetic Resonance (MR) images.
G. Charpiat, M. Hofmann, B. Schölkopf
Geometric Deformable Models
Abstract
Geometric deformable models are deformable models that are implemented using the level set method. They have been extensively studied and widely used in a variety of applications in biomedical image analysis. In this chapter, the general geometric deformable model framework is first presented and then recent developments on topology, prior shape, intensity and motion, resolution, efficiency, robust optimization, and multiple objects are reviewed. Key equations and motivating and demonstrative examples are provided for many methods and guidelines for appropriate use are noted.
Y. Bai, X. Han, J. L. Prince
Active Shape and Appearance Models
Abstract
Statistical models of shape and appearance are powerful tools for medical image analysis. The shape models can capture the mean and variation in shape of a structure or set of structures across a population. They can be used to help interpret new images by finding the parameters which best match an instance of the model to the image. Two widely used methods for matching are the Active Shape Model and the Active Appearance Model. We describe the models and the matching algorithms, and give examples of their use.
T. F. Cootes, M. G. Roberts, K. O. Babalola, C. J. Taylor

Statistical & Physiological Models

Frontmatter
Statistical Atlases
Abstract
This chapter discusses the general concept of statistical atlases built from medical images. A statistical atlas is a quantitative reflection of normal variability in anatomy, function, pathology, or other imaging measurements, and it allows us to establish a baseline against which abnormal images are to be compared for diagnostic or treatment planning purposes. Constructing a statistical atlas relies on a fundamental building block, namely deformable registration, which maps imaging data from many individuals to a common coordinate system, so that statistics of normal variability, as well as abnormal deviations from it, can be performed. 3D and 4D registration methods are discussed. This chapter also discusses the statistical analyses applied to co-registered normative images, and finally briefly touches upon use of machine learning for detection of imaging patterns that distinctly deviate from the normative range to allow for individualized classification.
C. Davatzikos, R. Verma, D. Shen
Statistical Computing on Non-Linear Spaces for Computational Anatomy
Abstract
Computational anatomy is an emerging discipline that aims at analyzing and modeling the individual anatomy of organs and their biological variability across a population. However, understanding and modeling the shape of organs is made difficult by the absence of physical models for comparing different subjects, the complexity of shapes, and the high number of degrees of freedom implied. Moreover, the geometric nature of the anatomical features usually extracted raises the need for statistics on objects like curves, surfaces and deformations that do not belong to standard Euclidean spaces. We explain in this chapter how the Riemannian structure can provide a powerful framework to build generic statistical computing tools. We show that few computational tools derive for each Riemannian metric can be used in practice as the basic atoms to build more complex generic algorithms such as interpolation, filtering and anisotropic diffusion on fields of geometric features. This computational framework is illustrated with the analysis of the shape of the scoliotic spine and the modeling of the brain variability from sulcal lines where the results suggest new anatomical findings.
X. Pennec, P. Fillard
Building Patient-Specific Physical and Physiological Computational Models from Medical Images
Abstract
We describe a hierarchy of computational models of the human body operating at the geometrical, physical and physiological levels. Those models can be coupled with medical images which play a crucial role in the diagnosis, planning, control and follow-up of therapy. In this paper, we discuss the issue of building patient-specific physical and physiological models from macroscopic observations extracted from medical images. We illustrate the topic of model personalization with concrete examples in brain shift modeling, hepatic surgery simulation, cardiac and tumor growth modeling. We conclude this article with scientific perspectives.
H. Delingette, N. Ayache
Constructing a Patient-Specific Model Heart from CT Data
Abstract
The goal of our work is to predict the patterns of blood flow in a model of the human heart using the Immersed Boundary method. In this method, fluid is moved by forces associated with the deformation of flexible boundaries which are immersed in, and interacting with, the fluid. In the present work the boundary is comprised of the muscular walls and valve leaflets of the heart. The method benefits by having an anatomically correct model of the heart. This report describes the construction of a model based on CT data from a particular individual, opening up the possibility of simulating interventions in an individual for clinical purposes.
D. M. McQueen, T. O’Donnell, B. E. Griffith, C. S. Peskin
Image-based haemodynamics simulation in intracranial aneurysms
Abstract
Image-based haemodynamics simulation is a computational technique that combines patient-specific vascular modeling from medical images with Computational Fluid Dynamics techniques to approximate the complex blood flow characteristics of healthy and diseased vessels. Advances in image quality, algorithmic sophistication and computing power are enabling the introduction of such technology not only as a biomedical research tool but also for clinical practice. In particular, the interaction between haemodynamical forces and arterial wall biology is believed to play an important role in the formation, growth and, eventually, rupture of intracranial aneurysms. Due to the absence of ground truth image modalities to measure blood flow, image-based haemodynamics simulation represents an attractive tool to provide insight into the haemodynamics characteristics of intracranial aneurysms. In this chapter, we provide an overview of the main components of this technique, illustrate recent efforts in its validation and sensitivity analysis and discuss preliminary clinical studies and future research directions.
A. G. Radaelli, H. Bogunović, M. C. Villa Uriol, J. R. Cebral, A. F. Frangi

Biomedical Perception

Frontmatter
Atlas-based Segmentation
Abstract
Image segmentation is a main task in many medical applications such as surgical or radiation therapy planning, automatic labelling of anatomical structures or morphological and morphometrical studies. Segmentation in medical imaging is however challenging because of problems linked to low contrast images, fuzzy object-contours, similar intensities with adjacent objects of interest, etc. Using prior knowledge can help in the segmentation task. A widely used method consists to extract this prior knowledge from a reference image often called atlas. We review in this chapter the existing approaches for atlas-based segmentation in medical imaging and we focus on those based on a volume registration method. We present the problem of using atlas information for pathological image analysis and we propose our solution for atlas-based segmentation in MR image of the brain when large space-occupying lesions are present. Finally, we present the new research directions that aim at overcome current limitations of atlas-based segmentation approaches based on registration only.
M. Bach Cuadra, V. Duay, J.-Ph. Thiran
Integration of Topological Constraints in Medical Image Segmentation
Abstract
Topology is a strong global constraint that can be useful in generating geometrically accurate segmentations of anatomical structures. Conversely, topological “defects” or departures from the true topology of a structure due to segmentation errors can greatly reduce the utility of anatomical models. In this chapter we cover methods for integrating topological constraints into segmentation procedures in order to generate geometrically accurate and topologically correct models, which is critical for many clinical and research applications.
F. Ségonne, B. Fischl
Monte Carlo Sampling for the Segmentation of Tubular Structures
Abstract
In this paper, we present a multiple hypotheses testing for the segmentation of tubular structures in medical imaging that addresses appearance (scanner artifacts, pathologies,…) and geometric (bifurcations) non-linearities. Our method represents vessels/tubular structures as sequences of state vectors (vessel cuts/cross-sections), which are described by the position of the corresponding plane, the center of the vessel in this plane and its radius. Thus, 3D segmentation consists in finding the optimal sequence of 2D planes normal to the vessel’s centerline. This sequence of planes is modeled by a probability density function (pdf for short) which is maximized with respect to the parameters of the state vector. Such a pdf is approximated in a non-parametric way, the Particle Filter approach, that is able to express multiple hypotheses (branches). Validation using ground truth from clinical experts and very promising experimental results for the segmentation of the coronaries demonstrates the potential of the proposed approach.
C. Florin, N. Paragios, J. Williams
Non-rigid registration using free-form deformations
Abstract
Free-form deformations are a powerful geometric modeling technique which can be used to represent complex 3D deformations. In recent years, free-form deformations have gained significant popularity in algorithms for the non-rigid registration of medical images. In this chapter we show how free-form deformations can be used in non-rigid registration to model complex local deformations of 3D organs. In particular, we discuss diffeomorphic and non-diffeomorphic representations of 3D deformation fields using free-form deformations as well as different penalty functions that can be used to constrain the deformation fields during the registration. We also show how free-form deformations can be used in combination with mutual information-based similarity metrics for the registration of mono-modal and multi-modal medical images. Finally, we discuss applications of registration techniques based on free-form deformations for the analysis of images of the breast, heart and brain as well as for segmentation and shape modelling.
D. Rueckert, P. Aljabar
Image registration using mutual information
Abstract
Different imaging modalities, such as CT, MRI and PET, are based on different physical principles and capture different and often complementary information. Many applications in clinical practice benefit from an integrated visualization and combined analysis of such multimodal images. In many applications it is also necessary to compare images acquired at a different time points, such as in the analysis of dynamic image sequences or of follow-up studies. Analysis of a single scene from multiple images assumes that the geometrical correspondence or registration between these images is known, such that anatomically identical points can be precisely identified and compared in each of the images. But reliable automated retrospective fusion or registration of multimodality images based on intrinsic image features is complicated by their different photometric properties, by the complexity of the scene and by the large variety of clinical applications. Maximization of mutual information of corresponding voxel intensities allows for fully automated registration of multimodality images without need for segmentation or user intervention, which makes it well suited for routine clinical use in a variety of applications.
F. Maes, D. Loeckx, D. Vandermeulen, P. Suetens
Physical Model Based Recovery of Displacement and Deformations from 3D Medical Images
Abstract
Estimating tissue displacement and deformation from time-varying medical images is a common problem in biomedical image analysis. For example, in order to better manage patients with ischemic heart disease, it would be useful to know their current extent of injury. This can be assessed by accurately tracking the motion of the left ventricle of the beating heart. Another example of this type of application is estimating the displacement of brain tissue during neurosurgery. The latter application is necessary because the presurgical planning for these delicate surgeries is based on images that may not accurately reflect the intraopertave brain (due to the action of gravity and other forces). In both examples, the tissue deformation cannot be measured directly. Instead, a sparse set of (potentially noisy) displacement estimates are extracted from acquired images. In this chapter, we explain how to use the physical properties of underlying organs or structures to guide such estimations of deformation, using neurosurgery and cardiac motion as example cases.
P. Yang, C. Delorenzo, X. Papademetris, J. S. Duncan
Graph-based Deformable Image Registration
Abstract
Deformable image registration is a field that has received considerable attention in the medical image analysis community. As a consequence, there is an important body of works that aims to tackle deformable registration. In this chapter we review one class of these techniques that use discrete optimization, and more specifically Markov Random Field models. We begin the chapter by explaining how one can formulate the deformable registration problem as a minimal cost graph problem where the nodes of the graph corresponds to the deformation grid, the graph connectivity encodes regularization constraints, and the labels correspond to 3D displacements. We then explain the use of discrete models in intensity-based volumetric registration. In the third section, we detail the use of Gabor-based attribute vectors in the context of discrete deformable registration, demonstrating the versatility of the graph-based models. In the last section of the chapter, the case of landmark-based registration is discussed. We first explain the discrete graphical model behind establishing landmark correspondences, and then continue to show how one can integrate it with the intensity-based model towards creating enhanced models that combine the best of both worlds.
A. Sotiras, Y. Ou, N. Paragios, C. Davatzikos

Clinical Biomarkers

Frontmatter
Cardiovascular Informatics
Abstract
As cardiac imaging technology advances, large amounts of imaging data are being produced which are not being mined sufficiently by current diagnostic tools for early detection and diagnosis of cardiovascular disease. We aim to develop a computational framework to mine cardiac imaging data and provide quantitative measures for developing a new risk assessment method. In this chapter, we present novel methods to quantify pericardial fat in non-contrast cardiac computed tomography images automatically, and to detect and quantify neovascularization in the coronary vessels using intra-vascular ultrasound imaging.
I. A. Kakadiaris, U. Kurkure, A. Bandekar, S. O’Malley, M. Naghavi
Rheumatoid Arthritis Quantification using Appearance Models
Abstract
Rheumatoid arthritis (RA) is a chronic disease that affects joints of the human skeleton. During therapy and during clinical trials, the accurate and precise measurement of the disease development is of crucial importance. Manual scoring frameworks exhibit high inter-reader variability and therefore constrain therapeutical monitoring or comparative evaluations during clinical trials.
In this chapter an automatic method for the quantification of rheumatoid arthritis is described. It is largely based on appearance models, and analyses a radiograph with regard to the two main indicators of RA progression: joint space width narrowing and erosions on the bones.
With the automatic approach a transition from global scoring methods that integrate over the entire anatomy, towards local measurements and the tracking of individual pathological changes becomes feasible. This is expected to improve both specificity and sensitivity of imaging biomarkers. It can improve therapy monitoring in particular if subtle changes occur, and can enhance the significance of clinical trials.
G. Langs, P. Peloschek, H. Bischof, F. Kainberger
Medical Image Processing for Analysis of Colon Motility
Abstract
A precise analysis and diagnosis of colon motility dysfunctions with current methods is almost unachievable. This makes it extremely difficult for the clinical experts to decide for the right intervention such as colon resection. The use of Cine MRI for visualizing the colon motility is a very promising technique. In addition, if image segmentation and qualitative motion analysis provide the necessary tools, it could provide the appropriate diagnostic solution. In this work we define necessary steps in the image processing chain to obtain clinical relevant measurements for a computer aided diagnosis of colon motility dysfunctions. For each step, we develop methods for an efficient handling of the MRI time sequences. There is need for compensating the breathing motion since no respiratory gating can be used during acquisition. We segment the colon using a graph-cuts approach in 2D over time for further analysis and visualization. The analysis of the large bowel motility is done by tracking the diameter of the colon during the propagation of the peristaltic wave. The main objective of this work is to automatize the assessment of clinical parameters which can be used to define a clinical index for motility pathologies.
N. Navab, B. Glocker, O. Kutter, S. M. Kirchhoff, M. Reiser
Segmentation of Diseased Livers: A 3D Refinement Approach
Abstract
Liver segmentation is the first data analysis step in computer-aided planning of liver tumor resections. For clinical applicability, the segmentation approach must be able to cope with the high variation in shape and gray-value appearance of the liver. In this article we present a novel segmentation scheme based on a true 3D segmentation refinement concept utilizing a hybrid desktop/virtual reality user interface. The method consists of two main stages. First, an initial segmentation is generated using graph cuts. Second, a segmentation refinement step allows to fix arbitrary segmentation errors. We demonstrate the robustness of our method on ten contrast enhanced liver CT scans and compare it to fifteen other methods. Our segmentation approach copes successfully with the high variation found in patient data sets and allows to produce a segmentation in a time-efficient manner.
R. Beichel, C. Bauer, A. Bornik, E. Sorantin, H. Bischof

Emerging Modalities & Domains

Intra and inter subject analyses of brain functional Magnetic Resonance Images (fMRI)
Abstract
This chapter proposes a review of the most prominent issues in analysing brain functional Magnetic Resonance data. It introduces the domain for readers with no or little knowledge in the field. The introduction places the context and orients the reader in the many questions put to the data, and summarizes the currently most commonly applied approach. The second section deals with intra subject data analysis, emphasizing hemodynamic response estimation issues. The third section describes current approaches and advances in analysing group data in a standard coordinate system. The last section proposes new spatial models for group analyses. Overall, the chapter gives a brief overview of the field and details some specific advances that are important for application studies in cognitive neurosciences.
J. B. Poline, P. Ciuciu, A. Roche, B. Thirion
Diffusion Tensor Estimation, Regularization and Classification
Abstract
In this chapter, we explore diffusion tensor estimation, regularization and classification. To this end, we introduce a variational method for joint estimation and regularization of diffusion tensor fields from noisy raw data as well as a Support Vector Machine (SVM) based classification framework.
In order to simultaneously estimate and regularize diffusion tensor fields from noisy observations, we integrate the classic quadratic data fidelity term derived from the Stejskal-Tanner equation with a new smoothness term leading to a convex objective function. The regularization term is based on the assumption that the signal can be reconstructed using a weighted average of observations on a local neighborhood. The weights measure the similarity between tensors and are computed directly from the diffusion images. We preserve the positive semi-definiteness constraint using a projected gradient descent.
The classification framework we consider in this chapter allows linear as well as non linear separation of diffusion tensors using kernels defined on the space of symmetric positive definite matrices. The kernels are derived from their counterparts on the statistical manifold of multivariate Gaussian distributions with zero mean or from distance substitution in the Gaussian Radial Basis Function (RBF) kernel. Experimental results on diffusion tensor images of the human skeletal muscle (calf) show the potential of our algorithms both in denoising and SVM-driven Markov random field segmentation.
R. Neji, N. Azzabou, G. Fleury, N. Paragios
From Local Q-Ball Estimation to Fibre Crossing Tractography
Abstract
Fibre crossing is an important problem for most existing diffusion tensor imaging (DTI) based tractography algorithms. To overcome limitations of DTI, high angular resolution diffusion imaging (HARDI) techniques such as q-ball imaging (QBI) have been introduced. The purpose of this chapter is to first give state of the art review of the existing local HARDI reconstruction techniques as well as the existing HARDI-based tractography algorithms. Then, we describe our analytical QBI solution to reconstruct the diffusion orientation distribution function (ODF) of water molecules and we propose a spherical deconvolution method to transform the diffusion ODF into a sharper fibre ODF. Finally, we propose a new deterministic and a new probabilistic algorithm based on this fibre ODF. We show that the diffusion ODF and fibre ODF can recover fibre crossing in simulated data, in a biological phantom and in real datasets. The fibre ODF improves angular resolution of QBI by more than 15∘ and greatly improves tractography results in regions of complex fibre crossing, fanning and branching.
M. Descoteaux, R. Deriche
Segmentation of Clustered Cells in Microscopy Images by Geometric PDEs and Level Sets
Abstract
With the huge amount of cell images produced in bio-imaging, automatic methods for segmentation are needed in order to evaluate the content of the images with respect to types of cells and their sizes. Traditional PDE-based methods using level-sets can perform automatic segmentation, but do not perform well on images with clustered cells containing sub-structures. We present two modifications for popular methods and show the improved results.
A. Kuijper, B. Heise, Y. Zhou, L. He, H. Wolinski, S. Kohlwein
Atlas-based whole-body registration in mice
Abstract
In this chapter, we present a fully automated approach for whole-body segmentation of mice in CT data, based on articulated skeleton registration. The method uses an anatomical animal atlas where position and degrees of freedom for each joint have been specified. Based on the registration result of the skeleton, a set of corresponding landmarks on bone and joint locations is used to derive further correspondences on surface representations of the lung and the skin. While atlas-based registration is applied to the former, a local geodesic shape context is employed for the latter. Subsequently, major organs are mapped from the atlas to the subject domain using Thin-Plate-Spline approximation, constrained by the landmarks on the skeleton, the lung and the skin. Accuracy and precision of the skeleton registration as well as organ approximation results in a follow-up study are demonstrated.
M. Baiker, J. Dijkstra, J. Milles, C. W. G. M. Löwik, B. P. F. Lelieveldt
Potential carotid atherosclerosis biomarkers based on ultrasound image analysis
Abstract
It has been shown that computerized analysis of ultrasound images of the carotid artery may provide quantitative disease biomarkers and can potentially serve as a ”second opinion” in the diagnosis of carotid atherosclerosis. Extending the findings of previous work on the subject, a set of methodologies are presented in this chapter, suitable for application on two-dimensional B-mode ultrasound images. More specifically, a Hough-Transform-based technique for automatic segmentation of the arterial wall allows the estimation of the intima-media thickness and the arterial distension waveform, two widely used determinants of arterial disease. Texture features extracted from Fourier-, wavelet-, and Gabor-filter-based methods can characterize symptomatic and asymptomatic atheromatous plaque. Finally, a methodology based on least-squares optical flow is proposed for the analysis and quantification of motion of the arterial wall.The suggested methodologies allow the extraction of useful biomarkers for the study of (a) the physiology of the arterial wall and (b) the mechanisms of carotid atherosclerosis.
S. Golemati, J. Stoitsis, K. S. Nikita
Metadaten
Titel
Handbook of Biomedical Imaging
herausgegeben von
Nikos Paragios
James Duncan
Nicholas Ayache
Copyright-Jahr
2015
Verlag
Springer US
Electronic ISBN
978-0-387-09749-7
Print ISBN
978-0-387-09748-0
DOI
https://doi.org/10.1007/978-0-387-09749-7

Premium Partner

    Bildnachweise