Computer Vision for Microscopy Image Analysis (Computer Vision and Pattern Recognition)

Are you a computer scientist working on image analysis? Are you a biologist seeking tools to process the microscopy data from image-based experiments? __Computer Vision for Microscopy Image Analysis__ provides a comprehensive and in-depth discussion of modern computer vision techniques, in particular deep learning, for microscopy image analysis that will advance your efforts. Progress in imaging techniques has enabled the acquisition of large volumes of microscopy data and made it possible to conduct large-scale, image-based experiments for biomedical discovery. The main challenge and bottleneck in such experiments is the conversion of "big visual data" into interpretable information. Visual analysis of large-scale microscopy data is a daunting task. Computer vision has the potential to automate this task. One key advantage is that computers perform analysis more reproducibly and less subjectively than human annotators. Moreover, high-throughput microscopy calls for effective and efficient techniques as there are not enough human resources to advance science by manual annotation. This book articulates the strong need for biologists and computer vision experts to collaborate to overcome the limits of human visual perception, and devotes a chapter each to the major steps in analyzing microscopy images, such as detection and segmentation, classification, tracking, and event detection. Front Cover Computer Vision for Microscopy Image Analysis Copyright Contents Contributors Preface Chapter 1: A biologist's perspective on computer vision 1. Thesis 2. Audience 3. Aim 4. Vision 5. Why biologists need computer vision experts 6. Why computer scientists need biologists 7. The limits of human visual perception from digital images 8. Quantitative phenotypic traits, high-content analysis 9. Different metrics for career advancement 10. The collaboration relationship 11. Biologists interacting with computer vision products 12. Current needs in biology 13. Conclusions and future perspectives References Chapter 2: Microscopy image formation, restoration, and segmentation 1. Introduction 2. Image formation 2.1. Phase contrast microscopy image formation model 2.2. DIC microscopy image formation model 3. Optics-based image restoration 3.1. Image restoration using a linear imaging model 3.2. Dictionary-based restoration 3.3. Cell-sensitive microscopy imaging 3.3.1. Estimate the cell-sensitive camera response function 3.3.2. Restoring the irradiance signal 4. Cell segmentation 4.1. Generation of phase-homogeneous superpixel 4.2. Semisupervised cell segmentation 4.3. Online correction of cell segmentation 5. Conclusion References Chapter 3: Detection and segmentation in microscopy images 1. Introduction 1.1. Cell detection and segmentation in optical microscopy images 1.1.1. Intensity thresholding and filtering 1.1.2. Region-based segmentation 1.1.3. Deformable models 1.1.4. Convolutional neural networks 1.2. Neuronal segmentation in electron microscopy images 1.2.1. Traditional filtering approaches 1.2.2. Supervised learning 1.2.3. Convolutional neural networks 2. Cell detection and segmentation in optical microscopy images 2.1. Learning to detect cells using nonoverlapping extremal regions 2.2. U-Net: Convolutional networks for biomedical image segmentation 2.3. Improving the robustness of convolutional networks to appearance variability in biomedical images 2.4. Locality sensitive deep learning for detection and classification of nuclei in routine colon cancer histology images 3. Segmentation of neuronal structures in EM images 3.1. Multiclass, multiscale series contextual model for image segmentation 3.2. Residual deconvolutional networks for brain electron microscopy image segmentation 4. Conclusion References Chapter 4: Visual feature representation in microscopy image classification 1. Introduction 1.1. Feature representation 1.2. Proposed approaches 2. Fisher vector representation 2.1. SIFT 2.2. DBN 2.3. CNN 3. Separation-guided dimension reduction 3.1. Method design 3.1.1. Learning of model parameters 3.1.2. Creation of training data Instance-instance constraint Instance-class constraint 3.2. Experimental setup 3.3. Results and discussion 4. Supervised intraembedding 4.1. Method design 4.2. Experimental setup 4.3. Results and discussion 5. Conclusions References Chapter 5: Cell tracking in time-lapse microscopy image sequences 1. Traditional cell tracking approaches 1.1. Model-based tracking 1.2. Tracking by detection 1.3. Joint segmentation and tracking 2. Deep learning-based cell tracking approaches 2.1. Power of deep learning 2.2. Deep learning for object tracking 2.3. Deep learning for cell tracking in time-lapse microscopy images 2.3.1. Deep learning-based cell detection/segmentation 2.3.2. Deep learning-based cell mitosis detection 2.3.3. Deep learning-based cell tracking 3. Metrics for evaluating and comparing cell tracking performance 4. A note on particle tracking 5. Future directions for computer vision-based cell tracking 5.1. Data is oxygen 5.2. Less dependency on labeled data 5.3. Interactive systems toward automation References Chapter 6: Mitosis detection in biomedical images 1. Mitosis process and the detection problem 1.1. Evaluation metric 2. Medical image for mitosis detection 2.1. Phase contrast microscopy image 2.2. Differential interference contrast microscopy image 2.3. Fluorescence microscopy image 2.4. Histological imaging 3. Mitosis detection approaches 3.1. Tracking-based methods 3.2. Tracking-free methods 3.2.1. Morphological descriptors 3.2.2. Texture feature Local binary patterns (LBPs) Gray-level cooccurrence matrix (GLCM) Scale-invariant feature transform (SIFT) 3.2.3. GIST 3.2.4. Sparse representation Dictionary learning Sparse decomposition 3.2.5. 3D filtering 3.2.6. Convolutional neural networks (CNNs) 3.2.7. Conclusion 3.3. Hybrid methods 3.3.1. Structure overview Image preconditioning Sequence segmentation 3.3.2. Hidden Markov model (HMM) 3.3.3. Conditional random field (CRF) 3.3.4. Hidden conditional random field (HCRF) 3.3.5. Event detection conditional random field (EDCRF) 3.3.6. Two-labeled hidden conditional random field (TL-HCRF) 3.3.7. Hidden conditional random field and semi-Markov model (HCRF and SMM) 3.3.8. Conclusion 3.4. Deep-learning methods 3.4.1. Two-stream CNN 3.4.2. 3-Dimensional convolutional network (C3D) for mitosis detection 3.4.3. Long short-term memory network (LSTM) for mitosis detection 4. Conclusion References Chapter 7: Object measurements from 2D microscopy images 1. Background 2. Introduction to 2D image measurements 3. Approach to numerical evaluations of feature variability and feature-based classification 4. Integration of open-source libraries for 2D image measurements 4.1. Integration of feature extraction libraries 4.2. Summary of features in all integrated libraries 5. Image features in scientific use cases 6. Variability of image features 6.1. Feature variability metric 6.2. Image feature variability analysis 6.3. Sources of image feature variations 6.3.1. Intensity features 6.3.2. Shape features 6.3.3. Textural Features 6.4. Discussion 7. Feature-based classification 7.1. Classification variability metric 7.2. Classification variability analysis 7.3. Discussion 8. Summary Appendix. Online information about the work References Chapter 8: Deep learning-based nuclei segmentation and classification in histopathology images with application to imaging ge 1. Joint nuclei segmentation and classification in histopathology images 1.1. Training data generation 1.1.1. Dataset 1.1.2. Preprocessing 1.2. Deep learning-based segmentation and classification 1.2.1. Network structure 1.2.2. Loss function 1.2.3. Postprocessing 1.2.4. Evaluation methods 1.3. Experimental results 2. Applications to imaging genomics 2.1. Intratumor heterogeneity 2.1.1. Intratumor spatial heterogeneity 2.1.2. Intratumor genetic heterogeneity 2.2. Tumor-Infiltrating Lymphocytes 3. Conclusion References Chapter 9: Open data and software for microscopy image analysis 1. Data is oxygen 2. Open data 3. Open software References Index Back Cover Are you a computer scientist working on image analysis? Are you a biologist seeking tools to process the microscopy data from image-based experiments? Computer Vision for Microscopy Image Analysis provides a comprehensive and in-depth discussion of modern computer vision techniques, in particular deep learning, for microscopy image analysis that will advance your efforts.Progress in imaging techniques has enabled the acquisition of large volumes of microscopy data and made it possible to conduct large-scale, image-based experiments for biomedical discovery. The main challenge and bottleneck in such experiments is the conversion of'big visual data'into interpretable information.Visual analysis of large-scale microscopy data is a daunting task. Computer vision has the potential to automate this task. One key advantage is that computers perform analysis more reproducibly and less subjectively than human annotators. Moreover, high-throughput microscopy calls for effective and efficient techniques as there are not enough human resources to advance science by manual annotation.This book articulates the strong need for biologists and computer vision experts to collaborate to overcome the limits of human visual perception, and devotes a chapter each to the major steps in analyzing microscopy images, such as detection and segmentation, classification, tracking, and event detection. Discover how computer vision can automate and enhance the human assessment of microscopy images for discovery Grasp the state-of-the-art approaches, especially deep neural networks Learn where to obtain open-source datasets and software to jumpstart his or her own investigation