Machine Learning and Medical Imaging

 
 
Academic Press
  • 1. Auflage
  • |
  • erschienen am 11. August 2016
  • |
  • 512 Seiten
 
E-Book | ePUB mit Adobe DRM | Systemvoraussetzungen
E-Book | PDF mit Adobe DRM | Systemvoraussetzungen
978-0-12-804114-7 (ISBN)
 

Machine Learning and Medical Imaging presents state-of- the-art machine learning methods in medical image analysis. It first summarizes cutting-edge machine learning algorithms in medical imaging, including not only classical probabilistic modeling and learning methods, but also recent breakthroughs in deep learning, sparse representation/coding, and big data hashing. In the second part leading research groups around the world present a wide spectrum of machine learning methods with application to different medical imaging modalities, clinical domains, and organs.

The biomedical imaging modalities include ultrasound, magnetic resonance imaging (MRI), computed tomography (CT), histology, and microscopy images. The targeted organs span the lung, liver, brain, and prostate, while there is also a treatment of examining genetic associations. Machine Learning and Medical Imaging is an ideal reference for medical imaging researchers, industry scientists and engineers, advanced undergraduate and graduate students, and clinicians.


  • Demonstrates the application of cutting-edge machine learning techniques to medical imaging problems
  • Covers an array of medical imaging applications including computer assisted diagnosis, image guided radiation therapy, landmark detection, imaging genomics, and brain connectomics
  • Features self-contained chapters with a thorough literature review
  • Assesses the development of future machine learning techniques and the further application of existing techniques
  • Englisch
  • San Diego
  • |
  • USA
Elsevier Science
  • 36,11 MB
978-0-12-804114-7 (9780128041147)
0128041145 (0128041145)
weitere Ausgaben werden ermittelt
  • Front Cover
  • Machine Learning and Medical Imaging
  • Copyright
  • Contents
  • Contributors
  • Editor Biographies
  • Preface
  • Acknowledgments
  • Part 1: Cutting-edge machine learning techniques in medical imaging
  • Chapter 1: Functional connectivity parcellation of the human brain
  • 1.1 Introduction
  • 1.2 Approaches to Connectivity-Based Brain Parcellation
  • 1.3 Mixture Model
  • 1.3.1 Model
  • 1.3.2 Inference
  • 1.4 Markov Random Field Model
  • 1.4.1 Model
  • 1.4.2 Inference
  • 1.5 Summary
  • References
  • Chapter 2: Kernel machine regression in neuroimaging genetics
  • 2.1 Introduction
  • 2.2 Mathematical Foundations
  • 2.2.1 From Regression Analysis to Kernel Methods
  • 2.2.2 Kernel Machine Regression
  • 2.2.3 Linear Mixed Effects Models
  • 2.2.4 Statistical Inference
  • 2.2.5 Constructing and Selecting Kernels
  • 2.2.6 Theoretical Extensions
  • 2.2.6.1 Generalized kernel machine regression
  • 2.2.6.2 Multiple kernel functions
  • 2.2.6.3 Correlated phenotypes
  • 2.2.6.4 Multidimensional traits
  • 2.3 Applications
  • 2.3.1 Genetic Association Studies
  • 2.3.2 Imaging Genetics
  • 2.4 Conclusion and Future Directions
  • Acknowledgments
  • Appendix A: Reproducing Kernel Hilbert Spaces
  • Appendix A.1: Inner Product and Hilbert Space
  • Appendix A.2: Kernel Function and Kernel Matrix
  • Appendix A.3: Reproducing Kernel Hilbert Space
  • Appendix A.4: Mercer's Theorem
  • Appendix A.5: Representer Theorem
  • Appendix B: Restricted Maximum Likelihood Estimation
  • References
  • Chapter 3: Deep learning of brain images and its application to multiple sclerosis
  • 3.1 Introduction
  • 3.1.1 Learning From Unlabeled Input Images
  • 3.1.1.1 From restricted Boltzmann machines to deep belief networks
  • Inference
  • Training
  • Deep belief networks
  • 3.1.1.2 Variants of restricted Boltzmann machines and deep belief networks
  • Convolutional DBNs
  • Alternative unit types
  • 3.1.1.3 Stacked denoising autoencoders
  • 3.1.2 Learning From Labeled Input Images
  • 3.1.2.1 Dense neural networks
  • 3.1.2.2 Convolutional neural networks
  • 3.2 Overview of Deep Learning in Neuroimaging
  • 3.2.1 Deformable Image Registration Using Deep-Learned Features
  • 3.2.2 Segmentation of Neuroimaging Data Using Deep Learning
  • 3.2.2.1 Hippocampus segmentation
  • 3.2.2.2 Infant brain image segmentation
  • 3.2.2.3 Brain tumor segmentation
  • 3.2.3 Classification of Neuroimaging Data Using Deep Learning
  • 3.2.3.1 Schizophrenia diagnosis
  • 3.2.3.2 Huntington disease diagnosis
  • 3.2.3.3 Task identification using functional MRI dataset
  • 3.2.3.4 Early diagnosis of Alzheimer's disease
  • 3.2.3.5 High-level 3D PET image feature learning
  • 3.3 Focus on Deep Learning in Multiple Sclerosis
  • 3.3.1 Multiple Sclerosis and the Role of Imaging
  • 3.3.2 White Matter Lesion Segmentation
  • 3.3.2.1 Patch-based segmentation methods
  • 3.3.2.2 Convolutional encoder network segmentation
  • 3.3.3 Modeling Disease Variability
  • 3.4 Future Research Needs
  • Acknowledgments
  • References
  • Chapter 4: Machine learning and its application in microscopic image analysis
  • 4.1 Introduction
  • 4.2 Detection
  • 4.2.1 Support Vector Machine
  • 4.2.1.1 Image preprocessing
  • 4.2.1.2 Robust ellipse fitting
  • 4.2.1.3 SVM-based ellipse refinement
  • Group 1
  • Group 2
  • Group 3
  • Group 4
  • Group 5
  • Group 6
  • 4.2.1.4 Inner geodesic distance-based clustering
  • 4.2.2 Deep Convolutional Neural Network
  • 4.2.2.1 CNN-based structured regression
  • 4.2.2.2 CNN architecture
  • 4.2.2.3 Structured prediction fusion and cell localization
  • 4.2.2.4 Experimental results
  • 4.3 Segmentation
  • 4.3.1 Random Forests
  • 4.3.1.1 Structured edge detection
  • 4.3.1.2 Hierarchical image segmentation
  • 4.3.1.3 Experimental results
  • 4.3.2 Sparsity-Based Dictionary Learning
  • 4.3.2.1 CNN-based shape initialization
  • 4.3.2.2 Sparsity-based shape modeling
  • 4.3.2.3 Local repulsive deformable model
  • 4.3.2.4 Experimental results
  • 4.4 Summary
  • References
  • Chapter 5: Sparse models for imaging genetics
  • 5.1 Introduction
  • 5.2 Basic Sparse Models
  • 5.3 Structured Sparse Models
  • 5.3.1 Group Lasso and Sparse Group Lasso
  • 5.3.2 Overlapping Group Lasso and Tree Lasso
  • 5.3.3 Fused Lasso and Graph Lasso
  • 5.4 Optimization Methods
  • 5.4.1 Proximal Gradient Descent
  • 5.4.2 Accelerated Gradient Method
  • 5.5 Screening
  • 5.5.1 Screening for Lasso
  • 5.5.1.1 Background
  • 5.5.1.2 Enhanced DPP (EDPP) screening rules
  • 5.5.1.3 Applications of EDPP to imaging genetics
  • 5.5.2 Screening Methods for Other Sparse Models
  • 5.6 Conclusions
  • References
  • Chapter 6: Dictionary learning for medical image denoising, reconstruction, and segmentation
  • 6.1 Introduction
  • 6.1.1 The Convenience of Orthogonal Transforms
  • 6.1.2 The Flexibility of Overcomplete Dictionaries
  • 6.2 Sparse Coding and Dictionary Learning
  • 6.2.1 Sparse Coding
  • 6.2.2 Dictionary Learning Problem
  • 6.2.3 K-SVD Dictionary Learning
  • 6.2.4 Online Dictionary Learning
  • 6.3 Patch-Based Dictionary Sparse Coding
  • 6.3.1 Overcompleteness
  • 6.3.2 Redundancy
  • 6.3.3 Adaptability
  • 6.4 Application of Dictionary Learning in Medical Imaging
  • 6.4.1 Denoising
  • 6.4.2 Reconstruction
  • 6.4.3 Super-Resolution
  • 6.4.4 Segmentation
  • 6.5 Future Directions
  • 6.6 Conclusion
  • References
  • Glossary
  • Chapter 7: Advanced sparsity techniques in magnetic resonance imaging
  • 7.1 Introduction
  • 7.2 Standard Sparsity in CS-MRI
  • 7.2.1 Model and Algorithm
  • 7.2.1.1 Related acceleration algorithm
  • 7.2.1.2 CSA and FCSA
  • Sketch Proof of Theorem 7.2
  • End of Proof
  • 7.2.2 Evaluation
  • 7.2.2.1 Experimental setup
  • 7.2.2.2 Visual comparisons
  • 7.2.2.3 CPU time and SNRs
  • 7.2.2.4 Sample ratios
  • 7.2.3 Summary
  • 7.3 Group Sparsity in Multicontrast MRI
  • 7.3.1 Model and Algorithm
  • 7.3.1.1 Proposed fast multicontrast reconstruction
  • 7.3.2 Evaluation
  • 7.3.2.1 SRI24 multichannel brain atlas data
  • 7.3.2.2 Complex-valued Shepp-Logan phantoms data
  • 7.3.2.3 Complex-valued turbo spin echo slices with early and late TEs data
  • 7.3.2.4 The benefit of group sparsity on both wavelet and gradient domains
  • 7.3.2.5 Results on SRI24 multichannel brain atlas data
  • 7.3.2.6 Results on Complex-valued Shepp-Logan phantoms data
  • 7.3.2.7 Results on Complex-valued turbo spin echo slices with early and late TEs
  • 7.3.2.8 Discussion
  • 7.3.3 Summary
  • 7.4 Tree Sparsity in Accelerated MRI
  • 7.4.1 Model and Algorithm
  • 7.4.1.1 Unconstrained tree-based MRI
  • 7.4.1.2 Constrained tree-based MRI
  • 7.4.2 Evaluation
  • 7.4.2.1 Experimental setup
  • 7.4.2.2 Group configuration for tree sparsity
  • 7.4.2.3 Visual comparisons
  • 7.4.2.4 SNRs and CPU time
  • 7.4.2.5 Sampling ratios
  • 7.4.2.6 Complex-valued image with radial sampling mask
  • 7.4.3 Summary
  • 7.5 Forest Sparsity in Multichannel CS-MRI
  • 7.5.1 Model and Algorithm
  • 7.5.2 Evaluation
  • 7.5.2.1 Multicontrast MRI
  • 7.5.2.2 Parallel MRI
  • 7.5.3 Summary
  • 7.6 Conclusion
  • References
  • Chapter 8: Hashing-based large-scale medical image retrieval for computer-aided diagnosis
  • 8.1 Introduction
  • 8.2 Related Work
  • 8.3 Supervised Hashing for Large-Scale Retrieval
  • 8.3.1 Overview of Scalable Image Retrieval Framework
  • 8.3.2 Kernelized and Supervised Hashing
  • 8.3.2.1 Hashing method
  • 8.3.2.2 Kernelized hashing
  • 8.3.2.3 Supervised hashing
  • 8.4 Results
  • 8.5 Discussion and Future Work
  • References
  • Part 2: Successful applications in medical imaging
  • Chapter 9: Multitemplate-based multiview learning for Alzheimer's disease diagnosis
  • 9.1 Background
  • 9.2 Multiview Feature Representation With MR Imaging
  • 9.2.1 Preprocessing
  • 9.2.2 Template Selection
  • 9.2.3 Registration and Quantification
  • 9.2.4 Feature Extraction
  • 9.2.4.1 Watershed segmentation
  • 9.2.4.2 Regional feature aggregation
  • 9.2.4.3 Anatomical analysis
  • 9.3 Multiview Learning Methods for AD Diagnosis
  • 9.3.1 Feature Filtering-Based Multiview Learning
  • 9.3.2 Maximum-Margin-Based Representation Learning
  • 9.3.3 View-Centralized Multiview Learning
  • Ensemble classification
  • 9.3.4 Relationship-Induced Multiview Learning
  • 9.4 Experiments
  • 9.4.1 Subjects
  • 9.4.2 Experimental Settings
  • 9.4.3 Results of Feature Filtering-Based Method for AD/MCI Diagnosis
  • 9.4.4 Results of Maximum-Margin-Based Learning for AD/MCI Diagnosis
  • 9.4.5 Results of View-Centralized Learning for AD/MCI Diagnosis
  • 9.4.6 Results of Relationship-Induced Learning for AD/MCI Diagnosis
  • 9.5 Summary
  • References
  • Chapter 10: Machine learning as a means toward precision diagnostics and prognostics
  • 10.1 Introduction
  • 10.2 Dimensionality Reduction
  • 10.2.1 Dimensionality Reduction Through Spatial Grouping
  • 10.2.2 Spatial Grouping of Structural MRI
  • 10.2.2.1 Spatial grouping of rs-fMRI
  • 10.2.3 Statistically Driven Dimensionality Reduction
  • 10.2.3.1 Statistically driven dimensionality reduction of structural MRI
  • 10.2.3.2 Statistically driven dimensionality reduction of functional MRI
  • 10.3 Model Interpretation: From Classification to Statistical Significance Maps
  • 10.4 Heterogeneity
  • 10.4.1 Generative Framework
  • 10.4.2 Discriminative Framework
  • 10.4.3 Generative Discriminative Framework
  • 10.5 Applications
  • 10.5.1 Individualized Diagnostic Indices Using MRI
  • 10.5.2 MRI-Based Diagnosis of AD: The SPARE-AD
  • 10.5.3 Individualized Early Predictions
  • 10.6 Conclusion
  • References
  • Chapter 11: Learning and predicting respiratory motion from 4D CT lung images
  • 11.1 Introduction
  • 11.2 3D/4D CT Lung Image Processing
  • 11.2.1 Lung Field and Vessel Segmentation
  • 11.2.1.1 Lung field extraction
  • 11.2.1.2 Vessel segmentation using geometric active contour models
  • 11.2.1.3 Vascularity-oriented level-set (VOLES)
  • 11.2.2 Serial Image Segmentation and Registration
  • 11.2.2.1 4D registration
  • 11.2.2.2 4D segmentation
  • 11.3 Extracting and Estimating Motion Patterns From 4D CT
  • 11.3.1 A Lung Motion Estimation Framework
  • 11.3.2 Motion Estimation Models
  • 11.3.2.1 PCA model
  • 11.3.2.2 Kernel-PCA model
  • 11.3.2.3 Motion prediction using LS-SVM
  • 11.3.3 Experiments
  • 11.4 An Example for Image-Guided Intervention
  • 11.4.1 CTF Guidance With Motion Compensation
  • 11.4.2 The CT-CTF Registration Algorithm
  • 11.4.3 Experiments
  • 11.5 Concluding Remarks
  • Acknowledgment
  • References
  • Chapter 12: Learning pathological deviations from a normal pattern of myocardial motion: Added value for CRT studies?
  • 12.1 Introduction
  • 12.1.1 Cardiac Resynchronization Therapy
  • 12.1.2 Patterns of Motion/Deformation
  • 12.1.3 Summary of the Challenges
  • 12.2 Features Extraction: Statistical Distance From Normal Motion
  • 12.2.1 Construction of Abnormality Maps
  • 12.2.2 Which Statistics for Motion Patterns?
  • 12.3 Manifold Learning: Characterizing Pathological Deviations From Normality
  • 12.3.1 Learning Part: Pathological Deviations From Normality (Manifold Learning)
  • 12.3.1.1 From high-dimensional motion patterns to low-dimensional coordinates (training set)
  • 12.3.1.2 Tuning parameters
  • 12.3.1.3 Visualization: data spread and main directions
  • 12.3.2 Testing Part: Distances to the Modeled Pathology and to Normality
  • 12.3.2.1 From high-dimensional motion patterns to low-dimensional coordinates (testing set)
  • 12.3.2.2 From low-dimensional motion coordinates to high-dimensional motion patterns (testing set)
  • 12.3.2.3 Projection to the manifold and distances computation
  • 12.3.2.4 Tuning parameters
  • 12.4 Back to the Clinical Application: Understanding CRT-Induced Changes
  • 12.4.1 Analysis Per Population
  • 12.4.2 Link With CRT Response
  • 12.5 Discussion/Future Work
  • 12.5.1 Pattern-Based Comparisons
  • 12.5.2 Going Beyond Parsai's Paper?
  • Acknowledgments
  • References
  • Chapter 13: From point to surface: Hierarchical parsing of human anatomy in medical images using machine learning technologie
  • 13.1 Introduction
  • 13.2 Literature Review
  • 13.3 Anatomy Landmark Detection
  • 13.3.1 Learning-Based Landmark Detection
  • 13.3.2 Application: Automatic Localize and Label of Vertebrae
  • 13.4 Detection of Anatomical Boxes
  • 13.4.1 Robust Anatomical Box Detection Using an Ensemble of Local Spatial Configurations
  • 13.4.2 Application: Autoalignment for MR Knee Scan Planning
  • 13.5 Coarse Organ Segmentation
  • 13.5.1 Sparse Shape Composition for Organ Localization
  • 13.5.2 Lung Region Localization in Chest X-Ray
  • 13.6 Precise Organ Segmentation
  • 13.6.1 Deformable Segmentation Using Hierarchical Clustering and Learning
  • 13.6.1.1 Affinity propagation clustering
  • 13.6.1.2 Iterative feature selection/clustering
  • 13.6.1.3 Learn boundary detectors
  • 13.6.2 Liver Segmentation in Whole-Body PET-CT Scans
  • 13.7 Conclusion
  • References
  • Chapter 14: Machine learning in brain imaging genomics
  • 14.1 Introduction
  • 14.2 Mining Imaging Genomic Associations Via Regression or Correlation Analysis
  • 14.2.1 Single-Locus Analysis
  • 14.2.1.1 Multiple comparison correction
  • 14.2.2 Multilocus Effects
  • 14.2.3 Multi-SNP-Multi-QT Associations
  • 14.3 Mining Higher Level Imaging Genomic Associations Via Set-Based Analysis
  • 14.3.1 Context-Based Test
  • 14.3.1.1 Over-representation analysis
  • 14.3.1.2 Rank-based enrichment analysis
  • 14.3.2 Context-Free Test
  • 14.3.3 Two-Dimensional Imaging Genomic Enrichment Analysis
  • 14.4 Discussion
  • 14.4.1 Prominent Findings
  • 14.4.2 Future Directions
  • References
  • Chapter 15: Holistic atlases of functional networks and interactions (HAFNI)
  • 15.1 Introduction
  • 15.2 HAFNI for Functional Brain Network Identification
  • 15.3 HAFNI Applications
  • 15.4 HAFNI-Based New Methods
  • 15.5 Future Directions of HAFNI Applications
  • Acknowledgments
  • References
  • Chapter 16: Neuronal network architecture and temporal lobe epilepsy: A connectome-based and machine learning study
  • 16.1 Introduction
  • 16.1.1 Treatment Outcome Prediction of Patients With TLE
  • 16.1.2 Naming Impairment Performance of Patients With TLE
  • 16.2 Treatment Outcome Prediction of Patients With TLE
  • 16.2.1 Participants
  • 16.2.2 Presurgical Image Acquisition and Processing
  • 16.2.3 Presurgical Connectome Reconstruction
  • 16.2.4 Connectome Prediction Framework
  • 16.2.4.1 Connectome feature selection component
  • 16.2.5 Results and Evaluation
  • 16.2.5.1 Stage-1 TLE prediction pipeline
  • 16.2.5.2 Stage-2 surgical treatment outcome prediction pipeline
  • 16.3 Naming Impairment Performance of Patients With TLE
  • 16.3.1 Participants
  • 16.3.2 Language Assessment
  • 16.3.3 Image Acquisition and Processing
  • 16.3.4 Connectome Reconstruction
  • 16.3.5 Connectome Prediction Framework
  • 16.3.6 Results and Evaluation
  • References
  • Index
  • Back Cover

Dateiformat: EPUB
Kopierschutz: Adobe-DRM (Digital Rights Management)

Systemvoraussetzungen:

Computer (Windows; MacOS X; Linux): Installieren Sie bereits vor dem Download die kostenlose Software Adobe Digital Editions (siehe E-Book Hilfe).

Tablet/Smartphone (Android; iOS): Installieren Sie bereits vor dem Download die kostenlose App Adobe Digital Editions (siehe E-Book Hilfe).

E-Book-Reader: Bookeen, Kobo, Pocketbook, Sony, Tolino u.v.a.m. (nicht Kindle)

Das Dateiformat EPUB ist sehr gut für Romane und Sachbücher geeignet - also für "fließenden" Text ohne komplexes Layout. Bei E-Readern oder Smartphones passt sich der Zeilen- und Seitenumbruch automatisch den kleinen Displays an. Mit Adobe-DRM wird hier ein "harter" Kopierschutz verwendet. Wenn die notwendigen Voraussetzungen nicht vorliegen, können Sie das E-Book leider nicht öffnen. Daher müssen Sie bereits vor dem Download Ihre Lese-Hardware vorbereiten.

Weitere Informationen finden Sie in unserer E-Book Hilfe.


Dateiformat: PDF
Kopierschutz: Adobe-DRM (Digital Rights Management)

Systemvoraussetzungen:

Computer (Windows; MacOS X; Linux): Installieren Sie bereits vor dem Download die kostenlose Software Adobe Digital Editions (siehe E-Book Hilfe).

Tablet/Smartphone (Android; iOS): Installieren Sie bereits vor dem Download die kostenlose App Adobe Digital Editions (siehe E-Book Hilfe).

E-Book-Reader: Bookeen, Kobo, Pocketbook, Sony, Tolino u.v.a.m. (nicht Kindle)

Das Dateiformat PDF zeigt auf jeder Hardware eine Buchseite stets identisch an. Daher ist eine PDF auch für ein komplexes Layout geeignet, wie es bei Lehr- und Fachbüchern verwendet wird (Bilder, Tabellen, Spalten, Fußnoten). Bei kleinen Displays von E-Readern oder Smartphones sind PDF leider eher nervig, weil zu viel Scrollen notwendig ist. Mit Adobe-DRM wird hier ein "harter" Kopierschutz verwendet. Wenn die notwendigen Voraussetzungen nicht vorliegen, können Sie das E-Book leider nicht öffnen. Daher müssen Sie bereits vor dem Download Ihre Lese-Hardware vorbereiten.

Weitere Informationen finden Sie in unserer E-Book Hilfe.


Download (sofort verfügbar)

107,04 €
inkl. 19% MwSt.
Download / Einzel-Lizenz
ePUB mit Adobe DRM
siehe Systemvoraussetzungen
PDF mit Adobe DRM
siehe Systemvoraussetzungen
Hinweis: Die Auswahl des von Ihnen gewünschten Dateiformats und des Kopierschutzes erfolgt erst im System des E-Book Anbieters
E-Book bestellen

Unsere Web-Seiten verwenden Cookies. Mit der Nutzung des WebShops erklären Sie sich damit einverstanden. Mehr Informationen finden Sie in unserem Datenschutzhinweis. Ok