Chapter
1 An Introduction to Neural Networks and Deep Learning
1.2 Feed-Forward Neural Networks
1.2.2 Multi-Layer Neural Network
1.2.3 Learning in Feed-Forward Neural Networks
1.3 Convolutional Neural Networks
1.3.1 Convolution and Pooling Layer
1.3.2 Computing Gradients
1.4.1 Vanishing Gradient Problem
1.4.2 Deep Neural Networks
1.4.3 Deep Generative Models
1.5 Tricks for Better Learning
1.5.1 Rectified Linear Unit (ReLU)
1.5.3 Batch Normalization
1.6 Open-Source Tools for Deep Learning
2 An Introduction to Deep Convolutional Neural Nets for Computer Vision
2.2 Convolutional Neural Networks
2.2.1 Building Blocks of CNNs
2.2.4 Tricks to Increase Performance
2.2.5 Putting It All Together: AlexNet
2.2.6 Using Pre-Trained CNNs
2.3.2 Fully Convolutional Networks
2.3.3 Multi-Modal Networks
2.3.5 Hybrid Learning Methods
2.4 Software for Deep Learning
Part 2 Medical Image Detection and Recognition
3 Efficient Medical Image Parsing
3.2 Background and Motivation
3.2.1 Object Localization and Segmentation: Challenges
3.3.1 Problem Formulation
3.3.2 Sparse Adaptive Deep Neural Networks
3.3.3 Marginal Space Deep Learning
3.3.4 An Artificial Agent for Image Parsing
3.4.1 Anatomy Detection and Segmentation in 3D
3.4.2 Landmark Detection in 2D and 3D
4 Multi-Instance Multi-Stage Deep Learning for Medical Image Recognition
4.3.1 Problem Statement and Framework Overview
4.3.2 Learning Stage I: Multi-Instance CNN Pre-Train
4.3.3 Learning Stage II: CNN Boosting
4.3.4 Run-Time Classification
4.4.1 Image Classification on Synthetic Data
4.4.2 Body-Part Recognition on CT Slices
4.5 Discussion and Future Work
5 Automatic Interpretation of Carotid Intima-Media Thickness Videos Using Convolutional Neural Networks
5.4.1 Convolutional Neural Networks (CNNs)
5.4.4 Intima-Media Thickness Measurement
5.5.1 Pre- and Post-Processing for Frame Selection
5.5.2 Constrained ROI Localization
5.5.3 Intima-Media Thickness Measurement
5.5.4 End-to-End CIMT Measurement
6 Deep Cascaded Networks for Sparsely Distributed Object Detection from Medical Images
6.2.1 Coarse Retrieval Model
6.2.2 Fine Discrimination Model
6.3 Mitosis Detection from Histology Images
6.3.2 Transfer Learning from Cross-Domain
6.3.3 Dataset and Preprocessing
6.3.4 Quantitative Evaluation and Comparison
6.4 Cerebral Microbleed Detection from MR Volumes
6.4.2 3D Cascaded Networks
6.4.3 Dataset and Preprocessing
6.4.4 Quantitative Evaluation and Comparison
6.4.5 System Implementation
6.5 Discussion and Conclusion
7 Deep Voting and Structured Regression for Microscopy Image Analysis
7.1 Deep Voting: A Robust Approach Toward Nucleus Localization in Microscopy Images
7.1.3 Weighted Voting Density Estimation
7.2 Structured Regression for Robust Cell Detection Using Convolutional Neural Network
7.2.3 Experimental Results
Part 3 Medical Image Segmentation
8 Deep Learning Tissue Segmentation in Cardiac Histopathology Images
8.2 Experimental Design and Implementation
8.2.1 Data Set Description
8.2.2 Manual Ground Truth Annotations
8.2.4 Training a Model Using Engineered Features
8.2.6 Testing and Performance Evaluation
8.3 Results and Discussion
8.3.1 Experiment 1: Comparison of Deep Learning and Random Forest Segmentation
8.3.2 Experiment 2: Evaluating the Sensitivity of Deep Learning to Training Data
9 Deformable MR Prostate Segmentation via Deep Feature Learning and Sparse Patch Matching
9.2.2 Learning Deep Feature Representation
9.2.3 Segmentation Using Learned Feature Representation
9.3.1 Evaluation of the Performance of Deep-Learned Features
9.3.2 Evaluation of the Performance of Deformable Model
10 Characterization of Errors in Deep Learning-Based Brain MRI Segmentation
10.2 Deep Learning for Segmentation
10.3 Convolutional Neural Network Architecture
10.3.1 Basic CNN Architecture
10.3.2 Tri-Planar CNN for 3D Image Analysis
10.4.4 Estimation of Centroid Distances
10.4.5 Registration-Based Segmentation
10.4.6 Characterization of Errors
10.5.1 Overall Performance
Part 4 Medical Image Registration
11 Scalable High Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning
11.2.2 Learn Intrinsic Feature Representations by Unsupervised Deep Learning
11.2.3 Registration Using Learned Feature Representations
11.3.1 Experimental Result on ADNI Dataset
11.3.2 Experimental Result on LONI Dataset
11.3.3 Experimental Result on 7.0-T MR Image Dataset
12 Convolutional Neural Networks for Robust and Real-Time 2-D/3-D Registration
12.4.1 Parameter Space Partitioning
12.4.2 Marginal Space Regression
12.5.1 Local Image Residual
12.5.2 3-D Points of Interest
12.6 Convolutional Neural Network
12.7 Experiments and Results
12.7.2 Hardware & Software
12.7.3 Performance Analysis
12.7.4 Comparison with State-of-the-Art Methods
Part 5 Computer-Aided Diagnosis and Disease Quantification
13 Chest Radiograph Pathology Categorization via Transfer Learning
13.2 Image Representation Schemes with Classical (Non-Deep) Features
13.2.1 Classical Filtering
13.2.2 Bag-of-Visual-Words Model
13.3 Extracting Deep Features from a Pre-Trained CNN Model
13.4 Extending the Representation Using Feature Fusion and Selection
13.5 Experiments and Results
13.5.2 Experimental Setup
13.5.3 Experimental Results
14 Deep Learning Models for Classifying Mammogram Exams Containing Unregistered Multi-View Images and Segmentation Maps of Lesions
14.3.1 Deep Learning Model
14.4 Materials and Methods
15 Randomized Deep Learning Methods for Clinical Trial Enrichment and Design in Alzheimer's Disease
15.2.1 Clinical Trials and Sample Enrichment
15.2.3 Backpropagation and Deep Learning
15.3 Optimal Enrichment Criterion
15.3.1 Ensemble Learning and Randomization
15.4 Randomized Deep Networks
15.4.1 Small Sample Regime and Multiple Modalities
15.4.3 RDA and RDR Training
15.4.4 The Disease Markers - RDAM and RDRM
15.5.1 Participant Data and Preprocessing
16 Deep Networks and Mutual Information Maximization for Cross-Modal Medical Image Synthesis
16.2 Supervised Synthesis Using Location-Sensitive Deep Network
16.2.2 Network Simplification
16.3 Unsupervised Synthesis Using Mutual Information Maximization
16.3.1 Generating Multiple Target Modality Candidates
16.3.2 Full Image Synthesis Using Best Candidates
16.3.3 Refinement Using Coupled Sparse Representation
16.3.4 Extension to Supervised Setting
16.4 Conclusions and Future Work
17 Natural Language Processing for Large-Scale Medical Image Analysis Using Deep Learning
17.2 Fundamentals of Natural Language Processing
17.3 Neural Language Models
17.3.2 Recurrent Language Model
17.4.1 UMLS Metathesaurus
17.5 Predicting Presence or Absence of Frequent Disease Types
17.5.1 Mining Presence/Absence of Frequent Disease Terms
17.5.2 Prediction Result and Discussion
Deep Learning for Medical Image Analysis
Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.
