UID:
edoccha_9960161211502883
Format:
1 online resource (417 pages).
ISBN:
0-12-812321-4
,
0-12-812133-5
Series Statement:
Elsevier and MICCAI Society Book Series
Note:
Front Cover -- Biomedical Texture Analysis -- Copyright -- Contents -- Preface -- 1 Fundamentals of Texture Processing for Biomedical Image Analysis -- 1.1 Introduction -- 1.2 Biomedical texture processes -- 1.2.1 Image intensity versus image texture -- 1.2.2 Notation and sampling -- 1.2.3 Texture functions as realizations of texture processes -- 1.2.3.1 Texture stationarity -- 1.2.4 Primitives and textons -- 1.2.5 Biomedical image modalities -- 1.3 Biomedical Texture Analysis (BTA) -- 1.3.1 Texture operators and aggregation functions -- 1.3.2 Normalization -- 1.3.3 Invariances -- 1.3.3.1 Invariance and equivariance of operators -- 1.3.3.2 Invariances of texture measurements -- 1.3.3.3 Nongeometric invariances -- 1.4 Conclusions -- Acknowledgments -- References -- 2 Multiscale and Multidirectional Biomedical Texture Analysis -- 2.1 Introduction -- 2.2 Notation -- 2.3 Multiscale image analysis -- 2.3.1 Spatial versus spectral coverage of linear operators: the uncertainty principle -- 2.3.2 Region of interest and response map aggregation -- 2.4 Multidirectional image analysis -- 2.4.1 The Local Organization of Image Directions (LOID) -- 2.4.2 Directional sensitivity of texture operators -- 2.4.3 Locally rotation-invariant operators and moving frames representations -- 2.4.4 Directionally insensitive, sensitive, and moving frames representations for texture classi cation: a quantitative performance comparison -- 2.5 Discussions and conclusions -- Acknowledgments -- References -- 3 Biomedical Texture Operators and Aggregation Functions -- 3.1 Introduction -- 3.2 Convolutional approaches -- 3.2.1 Circularly/spherically symmetric lters -- 3.2.2 Directional lters -- 3.2.2.1 Gabor wavelets -- 3.2.2.2 Maximum Response 8 (MR8) -- 3.2.2.3 Histogram of Oriented Gradients (HOG) -- 3.2.2.4 Riesz transform -- 3.2.3 Learned lters.
,
3.2.3.1 Steerable Wavelet Machines (SWM) -- 3.2.3.2 Dictionary Learning (DL) -- 3.2.3.3 Deep Convolutional Neural Networks (CNN) -- 3.2.3.4 Data augmentation -- 3.3 Gray-level matrices -- 3.3.1 Gray-Level Cooccurrence Matrices (GLCM) -- 3.3.2 Gray-Level Run-Length Matrices (GLRLM) -- 3.3.3 Gray-Level Size Zone Matrices (GLSZM) -- 3.4 Local Binary Patterns (LBP) -- 3.5 Fractals -- 3.6 Discussions and conclusions -- Acknowledgments -- References -- 4 Deep Learning in Texture Analysis and Its Application to Tissue Image Classi cation -- 4.1 Introduction -- 4.2 Introduction to convolutional neural networks -- 4.2.1 Neurons and nonlinearity -- 4.2.2 Neural network -- 4.2.3 Training -- 4.2.3.1 Forward pass -- 4.2.3.2 Error -- 4.2.3.3 Backpropagation of the error -- 4.2.3.4 Stochastic gradient descent -- 4.2.3.5 Weights initialization -- 4.2.3.6 Regularization -- 4.2.4 CNN -- 4.2.4.1 Main building blocks -- 4.2.4.2 CNN architectures -- 4.2.4.3 Visualization -- 4.3 Deep learning for texture analysis: literature review -- 4.3.1 Early work -- 4.3.2 Texture speci c CNNs -- 4.3.3 CNNs for biomedical texture classi cation -- 4.4 End-to-end texture CNN: proposed solution -- 4.4.1 Method -- 4.4.2 Experiments -- 4.4.2.1 Details of the network -- 4.4.2.2 Datasets -- 4.4.3 Results and discussions -- 4.4.3.1 Networks from scratch and pretrained -- 4.4.3.2 Networks depth analysis -- 4.4.3.3 Domain transferability and visualization -- 4.4.3.4 Results on larger images -- 4.4.4 Combining texture and shape analysis -- 4.5 Application to tissue images classi cation -- 4.5.1 State-of-the-art -- 4.5.2 Method -- 4.5.3 Datasets -- 4.5.3.1 AGEMAP -- 4.5.3.2 Lymphoma -- 4.5.4 Results and discussions -- 4.6 Conclusion -- References -- 5 Fractals for Biomedical Texture Analysis -- 5.1 Introduction -- 5.2 Tissue texture -- 5.3 Basic concepts of fractal geometry.
,
5.3.1 Self-similarity -- 5.3.2 Ordered hierarchy -- 5.3.3 Degree of irregularity -- 5.4 Methods for computing fractal dimensions -- 5.4.1 Differential Box-Counting (DBC) method -- 5.4.2 Fractional Brownian motion methods -- 5.4.2.1 Power spectrum -- 5.4.2.2 Variogram -- 5.4.3 Area-based methods -- 5.4.3.1 Triangular prism -- 5.4.3.2 Isarithm -- 5.4.3.3 Robust estimator -- 5.4.3.4 Epsilon-blanket -- 5.5 Types of fractals -- 5.6 FD parameter estimation optimization -- 5.7 Lacunarity analysis -- 5.7.1 Assessing image texture sparsity -- 5.7.2 Rotation-invariance -- 5.7.3 Clinical signi cance -- 5.8 Tumor tissue characterization -- 5.8.1 Surface roughness -- 5.8.2 Tumor fractal analysis -- 5.9 Considerations for fractal analysis -- 5.9.1 Choosing a suitable method for estimating the FD -- 5.9.2 Multivariate fractal analysis -- 5.9.3 Performing fractal or multifractal analysis -- 5.10 Conclusion -- References -- 6 Handling of Feature Space Complexity for Texture Analysis in Medical Images -- 6.1 Introduction -- 6.2 Applications of texture analysis -- 6.2.1 Lesion detection -- 6.2.2 Disease categorization -- 6.2.3 Image retrieval -- 6.3 Review of classi cation methods -- 6.3.1 Ensemble classi cation -- 6.3.2 Subcategorization -- 6.3.3 Sparse representation -- 6.4 Subcategory-based ensemble classi cation -- 6.4.1 Large Margin Local Estimate (LMLE) -- 6.4.1.1 Reference subcategorization -- 6.4.1.2 Local estimate generation -- 6.4.1.3 Large margin aggregation -- 6.4.2 Locally-constrained subcluster representation ensemble -- 6.4.2.1 Subcluster generation -- 6.4.2.2 Basis representation -- 6.4.2.3 Representation fusion -- 6.5 Experiments -- 6.5.1 Dataset and implementation -- 6.5.2 Results of patch classi cation -- 6.6 Conclusions -- References -- 7 Rigid Motion Invariant Classi cation of 3D Textures and Its Application to Hepatic Tumor Detection.
,
7.1 Introduction -- 7.2 Isotropic multiresolution analysis -- 7.3 Implementing rotations with compactly supported re nable functions -- 7.4 Connecting IMRA with stochastic texture models using Gaussian Markov Random Fields (GMRF) -- 7.5 Feature space for rotationally invariant texture discrimination -- 7.5.1 Self-distance for 3D textures -- 7.6 3D texture-based features -- 7.7 Conclusion -- References -- 8 An Introduction to Radiomics: An Evolving Cornerstone of Precision Medicine -- 8.1 Introduction -- 8.2 Background on cancer care -- 8.2.1 Biomarkers and cancer care -- 8.2.2 Limitations of response assessment process -- 8.2.3 Limitations of characterization process -- 8.2.4 Limitations of current biomarkers -- 8.3 The potential areas of radiomics utility -- 8.4 Work ow of radiomics -- 8.4.1 Image acquisition -- 8.4.2 Segmentation -- 8.4.3 Feature extraction -- 8.4.4 Analysis and validation -- 8.5 Examples of radiomics literature -- 8.6 Challenges of radiomics -- 8.7 Conclusions -- References -- 9 Deep Learning Techniques on Texture Analysis of Chest and Breast Images -- 9.1 Introduction -- 9.2 Computer-aided detection -- 9.2.1 Lung nodule detection in CT scans -- 9.2.2 Other detection problems for CT scans -- 9.3 Computer-aided diagnosis -- 9.3.1 Computer-aided diagnosis on breast lesions in ultrasound images -- 9.3.2 Computer-aided diagnosis on pulmonary nodules in CT scans -- 9.3.3 Other computer-aided diagnosis problems in pulmonary CT scans -- 9.4 Automatic mapping from image content to the semantic terms -- 9.4.1 Semantic mapping with the conventional pattern recognition paradigm -- 9.4.2 Deep learning for semantic mapping -- 9.5 Conclusion -- Acknowledgments -- References -- 10 Analysis of Histopathology Images -- 10.1 Histopathology imaging: a challenge for texture analysis -- 10.2 Traditional machine learning approaches.
,
10.2.1 Preprocessing -- 10.2.1.1 Staining normalization -- 10.2.1.2 Illumination normalization -- 10.2.2 Detection and segmentation of structures -- 10.2.2.1 Nuclei and cells -- 10.2.2.2 Glands -- 10.2.3 Feature extraction -- 10.2.3.1 Object-level features -- 10.2.3.2 Architectural features -- 10.2.3.3 Global and window based features -- 10.2.3.4 Multiresolution approaches -- 10.2.4 Feature selection and dimensionality reduction -- 10.2.4.1 Feature selection -- 10.2.4.2 Dimensionality reduction -- 10.2.5 Classi cation -- 10.3 Deep learning approaches -- 10.3.1 Supervised and unsupervised feature learning architectures -- 10.3.2 Deep convolutional neural networks -- 10.3.3 Deep learning approaches to histopathology image analysis -- 10.4 Histopathology challenges -- 10.4.1 Datasets -- 10.4.2 Tasks -- 10.4.3 Evaluation metrics -- 10.4.3.1 Mitosis detection -- 10.4.3.2 Image classi cation -- 10.4.3.3 Structure localization and/or segmentation -- 10.5 Detecting mitoses -- 10.6 Frame and whole slide image classi cation -- 10.7 Structure segmentation -- 10.8 Discussion and conclusions -- Acknowledgments -- References -- 11 MaZda - A Framework for Biomedical Image Texture Analysis and Data Exploration -- 11.1 Introduction -- 11.1.1 Related work -- 11.2 Texture analysis with MaZda -- 11.2.1 Overview of the image analysis work ows -- 11.2.2 Regions of interest -- 11.2.3 Feature extraction -- 11.2.3.1 Texture features -- 11.2.3.2 Color features -- 11.2.3.3 Morphological features -- 11.2.4 Image preprocessing -- 11.2.5 Feature naming convention -- 11.2.6 Feature maps and image segmentation -- 11.2.7 Machine learning -- 11.2.7.1 Feature selection -- 11.2.7.2 Classi cation -- 11.2.7.3 Processing beyond MaZda - data mining using Weka -- 11.3 Applications -- 11.3.1 Lesion detection with MaZda and Weka -- 11.4 Summary -- Appendix 11.A.
,
11.A.1 List of feature name symbols.
Language:
English
Bookmarklink
