Deep Learning for Biomedical Data Analysis: Techniques, Approaches, and Applications

✍ Scribed by Mourad Elloumi

• Publisher: Springer
• Year: 2021
• Tongue: English
• Leaves: 358
• Edition: 1
• Category: Library

✦ Synopsis

This book is the first overview on Deep Learning (DL) for biomedical data analysis. It surveys the most recent techniques and approaches in this field, with both a broad coverage and enough depth to be of practical use to working professionals. This book offers enough fundamental and technical information on these techniques, approaches and the related problems without overcrowding the reader's head. It presents the results of the latest investigations in the field of DL for biomedical data analysis. The techniques and approaches presented in this book deal with the most important and/or the newest topics encountered in this field. They combine fundamental theory of Artificial Intelligence (AI), Machine Learning (ML) and DL with practical applications in Biology and Medicine. Certainly, the list of topics covered in this book is not exhaustive but these topics will shed light on the implications of the presented techniques and approaches on other topics in biomedical data analysis. The book finds a balance between theoretical and practical coverage of a wide range of issues in the field of biomedical data analysis, thanks to DL. The few published books on DL for biomedical data analysis either focus on specific topics or lack technical depth. The chapters presented in this book were selected for quality and relevance. The book also presents experiments that provide qualitative and quantitative overviews in the field of biomedical data analysis. 

The reader will require some familiarity with AI, ML and DL and will learn about techniques and approaches that deal with the most important and/or the newest topics encountered in the field of DL for biomedical data analysis. He/she will discover both the fundamentals behind DL techniques and approaches, and their applications on biomedical data. This book can also serve as a reference book for graduate courses in Bioinformatics, AI, ML and DL. The book aims not only at professional researchers and practitioners but also graduate students, senior undergraduate students and young researchers. This book will certainly show the way to new techniques and approaches to make new discoveries.

✦ Table of Contents

Contents
Part I Deep Learning for Biomedical Data Analysis
1-Dimensional Convolution Neural Network Classification Technique for Gene Expression Data
1 Introduction
2 Related Works
3 Preliminaries
3.1 Forward Selection and Minimum Redundancy: Maximum Relevance Criterion Filter Approach
3.2 Fisher Criterion Filter Approach
3.3 k-Means and Signal-to-Noise Ratio Filter Approach
3.3.1 k-Means Algorithm
3.3.2 Signal-to-Noise-Ratio Ranking Approach
3.4 Deep Learning
3.5 Convolutional Neural Network
3.5.1 2D Convolutional Neural Networks
3.5.2 1D Convolutional Neural Networks
4 Dataset Details
5 Proposed Approach
5.1 Preprocessing Using Fisher Ranking Approach
5.2 Model Creation
5.3 Model Training
5.4 Model Testing
6 Experimental Results
7 Conclusion
References
Classification of Sequences with Deep Artificial Neural Networks: Representation and Architectural Issues
1 Introduction
2 Sequences Representation
2.1 Representation of Fixed-Length Sequences
2.2 Spectral Representations
2.3 Frequency Chaos-Game Representations
3 Deep Learning Architectures for Sequences Classification
3.1 Restricted Boltzmann Machines Layers in Artificial Neural Networks
3.2 Convolutional Layers in Artificial Neural Networks
3.3 Recurrent Layers for Sequence Classification
3.4 Other Useful Layers in Neural Networks
3.5 Layers Assembly in a Deep Neural Network
4 Experiments and Results
4.1 Prediction of Nucleosomes
4.2 Bacteria Classification Using 16S Gene Sequences
4.3 Discussions
4.4 Execution Times
5 Conclusions
References
A Deep Learning Model for MicroRNA-Target Binding
1 Introduction
2 Related Work
3 Materials and Methods
3.1 Alignment and Representation Layer
3.2 Embedding Layer
3.3 Dropout Layer
3.4 LSTM Layer
3.5 Dense Layer
3.6 Optimization and Other Parameters
3.7 Implementation Details
4 Results
4.1 Datasets
4.2 Empirical Results
5 Conclusion
References
Recurrent Neural Networks Architectures for Accidental Fall Detection on Wearable Embedded Devices
1 Introduction
2 Related Works
2.1 Fall Detection Techniques
2.2 Datasets
3 Design
3.1 Dataset and Labeling
3.2 RNN Architecture
3.3 Training and Inference
4 Implementation and Results
4.1 Annotation Procedure
4.2 Software Implementation and Training
4.3 Comparison with Statistical Indicators
4.4 Embedded
5 Conclusions and Future Work
References
Part II Deep Learning for BiomedicalImage Analysis
Medical Image Retrieval System Using Deep Learning Techniques
1 Introduction
2 Image Retrieval Systems
2.1 Text-Based Image Retrieval
2.2 Content-Based Image Retrieval
2.2.1 Single Feature Based CBIR Systems
2.2.2 Combined Feature Based CBIR Systems
2.2.3 Hierarchical Feature Based CBIR Systems
2.3 Semantic-Based Image Retrieval
2.3.1 Machine Learning Based CBIR
2.3.2 Deep Learning Based CBIR
3 Applications of CBIR/SBIR
4 Deep Learning Based Medical Image Retrieval System
4.1 Deep Learning Models
4.2 Deep Learning Based Skin Cancer Image Retrieval
4.2.1 Survey on Skin Lesions
4.2.2 Dataset
4.2.3 Data Augmentation
4.2.4 Retrieval Using Inception V3 Model
4.2.5 Retrieval Using Inception ResNet V2 Model
4.2.6 Retrieval Using VGG-16 Model
4.2.7 Performance Analysis: Why VGG-16 Outperforms
4.2.8 Retrieval Using Modified VGG-16 Model
5 Conclusions
References
Medical Image Fusion Using Deep Learning
1 Introduction
2 Applications of Image Fusion Techniques
2.1 Fusion in Medical Imaging
2.2 Fusion in Detection and Tracking
2.3 Fusion in Recognition
2.4 Fusion in Color Vision
2.5 Fusion in Remote Sensing
2.6 Fusion in Surveillance
3 Image Fusion Approaches
3.1 Pixel Level
3.2 Feature Level
3.3 Decision Level
4 The General Medical Image Fusion Procedure
5 Major Medical Image Fusion Methods
6 Deep Learning Based Medical Image Fusion
6.1 Training
6.2 Fusion Strategy
7 Experimental Results and Discussion
8 Conclusion
References
Deep Learning for Histopathological Image Analysis
1 Introduction
2 Current Deep Learning Models for Digital Pathology
2.1 What Is Deep Learning?
2.2 Deep Learning for Classification
2.2.1 Detection
2.2.2 Scoring
2.2.3 Tissue Classification
2.3 Deep Learning for Segmentation
2.3.1 Cell Segmentation
2.3.2 Large Regions of Interest or Composite Objects Segmentation
3 Challenges and Opportunities
3.1 Annotations
3.2 Multiple Stainings
3.3 Generative Adversarial Network
4 Conclusion
References
Innovative Deep Learning Approach for Biomedical Data Instantiation and Visualization
1 Introduction
2 Deep Neural Networks: A Brief Introduction
2.1 From Shallow to Deep Neural Networks
2.2 Some Weaknesses
2.2.1 Finding the Best Neural Structure
2.2.2 The Interpretability of the Obtained Results
3 Emergent Architectures: Generative Networks
3.1 The Generative Adversarial Networks
3.2 Variational Auto-Encoders
4 Variational Autoencoders for Data Visualization and Analysis
4.1 Massive Amount of Unlabeled Data
4.2 Visual Support for Data Labeling
4.3 Identification and Reduction of the Conflict Area
5 Case Study: The Breast Cancer Wisconsin Dataset
5.1 Description of the Dataset
5.2 Evaluation Metrics
5.3 The Variational Auto-Encoder Used as Visual Support
5.4 Visualization of the Latent Space
5.5 Selection of the Training Samples from the 2D-Latent Space
5.5.1 Step1: Selection of the First Training Set
5.5.2 Step2: Identification of the Conflict Area
5.5.3 Step3: Reduction of the Conflict Area
5.5.4 Results Analysis
5.6 Random Selection of Training Samples
6 Conclusion
References
Convolutional Neural Networks in Advanced Biomedical Imaging Applications
1 Introduction
2 Scientific Foundations
2.1 Artificial Neural Networks
2.1.1 Basics of Artificial Neural Networks
2.1.2 From Feed Forward Artificial Neural Networks to Convolutional Neural Networks
2.1.3 Transfer Learning and Algorithm Interpretability
2.1.4 Relevant Terms
2.2 Advanced Optical Imaging Tools
2.2.1 Optical Coherence Tomography
2.2.2 Multiphoton Fluorescence Microscopy
2.2.3 Second Harmonic Generation
2.2.4 Coherent Raman Imaging
2.2.5 Advanced Imaging Analysis Challenges and Opportunities
3 Interdisciplinary Research Applications Involving Advanced Imaging and Deep Learning
3.1 Applied Optical Coherence Tomography Paired with Deep Learning Analysis
3.1.1 Analysis of Retinal Degeneration in Disease States
3.1.2 Segmenting Clinically Relevant Retinal Layers in OCT Images
3.1.3 CNN Analysis of Non-Retinal OCT Volumes
3.2 Non-linear Microscopy with CNN Analysis
3.2.1 CARS and CNNs for Automated Lung Cancer Diagnosis
3.2.2 CNN and Multiphoton Microscopy for Automated Ovarian Tissue Analysis
3.2.3 Summary
3.3 U-Nets and Advanced Microscopy Applications
3.3.1 U-Net-Inspired Algorithms for Image Segmentation
3.3.2 U-Net for Image Denoising
4 Data Management for CNNs in Imaging Applications
4.1 Case Studies
4.1.1 Detection of Diabetic Retinopathy in Retinal Fundus Photographs
4.1.2 Melanoma Diagnosis from Skin Photographs
4.1.3 Macular Disease Detection from OCT Scans
4.1.4 Cervical Cancer Diagnosis by Coherence Anti-Stokes Raman Scatttering
4.2 Building Systems with Small Training Datasets
4.3 Data Management Summary
5 Conclusions
References
Part III Deep Learning for Medical Diagnostics
Deep Learning for Lung Disease Detection from Chest X-Rays Images
1 Introduction
2 Classification of Deep Learning Techniques for Lung Disease Detection from Chest X-Rays Images
3 Deep Learning for the Thoracic Disease Detection
3.1 Datasets for the Thoracic Disease Detection
3.1.1 ChestX-ray8 Dataset
3.1.2 Indiana Dataset
3.2 Performance Metrics
3.3 Deep Learning Techniques for Thoracic Disease Detection
4 Deep Learning for Tuberculosis Detection
4.1 Datasets for Tuberculosis Detection
4.1.1 The Montgomery County Dataset
4.1.2 The Shenzen Dataset
4.1.3 The Belarus Dataset
4.2 Performance Metrics for Tuberculosis Detection
4.3 DL Techniques for Tuberculosis Detection
5 Deep Learning for Lung Nodule Detection
5.1 Datasets for Lung Nodule Detection
5.2 Performance Metrics for Lung Nodule Detection
5.3 Deep Learning Techniques for Lung Nodule Detection
6 Conclusion and Future Scope
References
Deep Learning in Multi-Omics Data Integration in Cancer Diagnostic
1 Introduction
2 Deep Learning for Multi-Omics Data Integration Methods Categories
2.1 Early Layer Integration
2.2 Middle Layer Integration
2.3 Late Layer Integration
3 Breast Cancer
4 Colorectal Cancer
5 Bladder Cancer
6 Conclusion
References
Using Deep Learning with Canadian Primary Care Data for Disease Diagnosis
1 Introduction
2 Extraction and Processing of the EMR Data at CPCSSN for Research
2.1 EMR Data Types
2.2 Challenges
2.3 Data Extraction, Anonymization, Transfer and Load
3 Artificial Intelligence in Medical Decision Support Systems
3.1 Types of ML Models
3.2 ML Algorithms
4 State-of-the-Art DL Research in Healthcare
5 Example Scenarios of Applications of DL to EMRs
5.1 Data Preprocessing
5.1.1 Processing Structured and Semi-Structured EMRs
5.1.2 Processing Unstructured Text Data
5.2 Model Validation Approaches
5.3 Research Studies
5.3.1 Predicting Hypertension Using Structured EMR Data
5.3.2 Diagnosing PTSD Using Structured EMR Data
5.3.3 Classification of Low Back Pain Using EMR Chart Note Data
5.3.4 Predicting COPD from EMR Data
5.3.5 Weak Supervision Model for Text Embedding to Diagnose PTSD
5.3.6 PTSD Diagnosis Using Encounter Notes and Deep CNN
5.3.7 Hybrid Models for Diagnosing PTSD
5.3.8 Other DL Models from State-of-the-Art Research on MDSS
6 Conclusion
References
Brain Tumor Segmentation and Surveillance with Deep Artificial Neural Networks
1 Introduction
2 Theoretical Background of the Problem
2.1 Image Segmentation
2.2 Brain Tumor Segmentation
2.3 Tumor Surveillance
2.4 Deep Learning Segmentation Task
2.5 Motivation
2.6 Challenges
3 Brain Tumor Segmentation Using Deep Artificial Neural Networks
3.1 Image Segmentation in Computer Vision Realm
3.2 Deep Artificial Neural Networks and Image Segmentation
3.3 DL-Based Image Segmentation Architectures
3.3.1 Convolutional Neural Networks
3.3.2 Fully Convolutional Networks
3.3.3 Encoder-Decoder Based Models
3.3.4 Other Deep Learning Models Used in Image Segmentation
3.4 Brain Tumor Segmentation Task Challenge
4 Inception Modules in Brain Tumor Segmentation
4.1 Brain Tumor Segmentation Using Inception and Dilated Inception modules
4.2 BraTS Dataset and Pre-processing
4.3 Deep Artificial Neural Network Architectures
4.3.1 Inception Module
4.3.2 Dilated Inception U-Net
4.3.3 Modified DSC as Objective/Loss Function
4.4 Experimental Setup and Results
4.4.1 Results from Inception Modules
4.4.2 Results from DIU-Net
5 Uncertainty Estimation in Brain Tumor Segmentation
5.1 Variational Learning
5.2 Variational Density Propagation
5.3 Extended Variational Density Propagation
5.3.1 First Convolutional Layer
5.3.2 Non-linear Activation Function
5.3.3 Max-Pooling Layer
5.3.4 Flattening Operation
5.3.5 Fully-Connected Layer
5.3.6 Softmax Function
5.3.7 Objective Function
5.3.8 Back-Propagation
5.4 Application to Brain Tumor Segmentation in MRI Images
6 Tumor Surveillance
6.1 Rationale for Tumor Surveillance
6.2 Surveillance Techniques
6.3 Community-Level Active Surveillance
6.4 Surveillance of Brain Tumor
6.4.1 An Example of Surveillance Study
7 Conclusion
References
Index