Predictive Intelligence in Medicine: Third International Workshop, PRIME 2020, Held in Conjunction with MICCAI 2020, Lima, Peru, October 8, 2020, Proceedings ... Recognition, and Graphics Book 12329)

This book constitutes the proceedings of the Second International Workshop on Predictive Intelligence in Medicine, PRIME 2020, held in conjunction with MICCAI 2020, in Lima, Peru, in October 2020. The workshop was held virtually due to the COVID-19 pandemic. The 17 full and 2 short papers presented in this volume were carefully reviewed and selected for inclusion in this book. The contributions describe new cutting-edge predictive models and methods that solve challenging problems in the medical field for a high-precision predictive medicine. Preface 6 Organization 9 Contents 10 Context-Aware Synergetic Multiplex Network for Multi-organ Segmentation of Cervical Cancer MRI 12 1 Introduction 13 2 Proposed Context-Aware Synergetic Multiplex Network for Multi-organ Segmentation 15 3 Results and Discussion 17 4 Conclusion 20 References 21 Residual Embedding Similarity-Based Network Selection for Predicting Brain Network Evolution Trajectory from a Single Observation 23 1 Introduction 24 2 Proposed Method 26 3 Results and Discussion 31 4 Conclusion 32 References 33 Adversarial Brain Multiplex Prediction from a Single Network for High-Order Connectional Gender-Specific Brain Mapping 35 1 Introduction 36 2 Proposed Deep Brain Multiplex Prediction Using G-GAN for Gender Fingerprinting 37 3 Results and Discussion 41 4 Conclusion 43 References 44 Learned Deep Radiomics for Survival Analysis with Attention 46 1 Introduction 47 2 Related Work 47 3 Method 48 3.1 From Cox Survival Model to Loss Function 49 3.2 CNN Model for Survival Analysis 49 4 Experimental Validation 52 4.1 Experiment 1. CNNs vs Classical Methods for Survival Analysis 53 4.2 Experiment 2. Evaluation of Deep Learning Methods 53 5 Discussion and Conclusions 54 References 55 Robustification of Segmentation Models Against Adversarial Perturbations in Medical Imaging 57 1 Introduction 57 2 Methodology 59 3 Experiments 63 4 Conclusion 67 References 67 Joint Clinical Data and CT Image Based Prognosis: A Case Study on Postoperative Pulmonary Venous Obstruction Prediction 69 1 Introduction 69 2 Methods 71 2.1 Clinical Data Based Method 71 2.2 CT Image Based Method 72 2.3 Joint Data and Image Based Method 73 3 Dataset 73 4 Preprocessing and Learning Techniques 74 4.1 Image Augmentation 74 4.2 Resampling 74 4.3 Loss Function Modification 75 5 Experiments 75 6 Conclusion 77 References 77 Low-Dose CT Denoising Using Octave Convolution with High and Low Frequency Bands 79 1 Introduction 79 1.1 Related Work 81 2 Method 82 3 Experimental Results 84 3.1 Dataset and Experimental Settings 84 3.2 Quantitative Results 86 3.3 Qualitative Results 86 4 Conclusions 88 References 88 Conditional Generative Adversarial Network for Predicting 3D Medical Images Affected by Alzheimer's Diseases 90 1 Introduction 90 2 Method 92 2.1 Details of Proposed Network 93 2.2 Objective Function 93 3 Experimental Results 94 4 Conclusion 99 References 99 Inpainting Cropped Diffusion MRI Using Deep Generative Models 102 1 Introduction 102 2 Methods 104 2.1 Model Architecture 104 2.2 Dataset and Evaluation Metrics 105 3 Results and Discussion 107 3.1 Evaluation on Artificially Cropped B0 MRIs of DWI 107 3.2 Impact on Downstream Preprocessing 108 4 Conclusion 109 References 109 Multi-view Brain HyperConnectome AutoEncoder for Brain State Classification 112 1 Introduction 113 2 Proposed HyperConnectome AutoEncoder (HCAE) for Brain State Classification 114 3 Results and Discussion 117 4 Conclusion 119 References 120 Foreseeing Brain Graph Evolution over Time Using Deep Adversarial Network Normalizer 122 1 Introduction 123 2 Proposed Method 124 3 Results and Discussion 130 4 Conclusion 132 References 132 Longitudinal Prediction of Radiation-Induced Anatomical Changes of Parotid Glands During Radiotherapy Using Deep Learning 134 1 Introduction 134 2 Related Works 135 3 Materials and Methods 136 3.1 Dataset 136 3.2 Deep Learning Model 137 3.3 Evaluation 138 4 Results 138 5 Discussion 140 6 Conclusion 141 References 141 Deep Parametric Mixtures for Modeling the Functional Connectome 144 1 Introduction 144 2 Methods 145 2.1 Deep Parametric Mixtures 145 2.2 Selecting Optimal Basis Y 146 3 Experiments 148 3.1 Data Pre-processing 148 3.2 Synthetic Experiments 149 4 Connectomes Specific to Sex, HIV, and Alcohol 150 5 Conclusion 152 References 152 Deep EvoGraphNet Architecture for Time-Dependent Brain Graph Data Synthesis from a Single Timepoint 155 1 Introduction 156 2 Proposed Method 157 3 Results and Discussion 162 4 Conclusion 164 References 164 Uniformizing Techniques to Process CT Scans with 3D CNNs for Tuberculosis Prediction 167 1 Introduction 167 2 Related Work 168 3 Methods 169 3.1 Uniformizing Techniques 169 3.2 Three-Dimensional (3D) CNN Architecture 171 4 Experiments 172 4.1 Dataset 172 4.2 Baseline and Implementation Details 172 4.3 Metrics 173 4.4 Results 173 5 Discussion and Conclusion 176 References 176 mr2NST: Multi-resolution and Multi-reference Neural Style Transfer for Mammography 180 1 Introduction 180 2 Method 182 2.1 Neural Style Transfer 182 2.2 Multiple Reference Style Images 183 3 Experiments and Results 184 4 Conclusion 188 References 188 Template-Oriented Multi-task Sparse Low-Rank Learning for Parkinson’s Diseases Diagnosis 189 1 Introduction 189 2 Method 191 2.1 Proposed Model 191 2.2 Optimization 193 3 Experiments 193 3.1 Experimental Setting 193 3.2 Data Preprocessing 194 3.3 Classification Performance for PD 194 3.4 Potential Disease-Related Brain Regions 195 4 Conclusion 197 References 197 Multimodal Prediction of Breast Cancer Relapse Prior to Neoadjuvant Chemotherapy Treatment 199 1 Introduction 199 2 Related Work 200 3 Methods 202 3.1 Dataset and Annotations 202 3.2 MRI Model 203 3.3 Clinical Model 205 3.4 Ensemble Model 205 4 Results 206 5 Discussion and Conclusion 208 References 209 A Self-ensembling Framework for Semi-supervised Knee Cartilage Defects Assessment with Dual-Consistency 211 1 Introduction 211 2 Methodology 213 2.1 Mean Teacher Mechanism 213 2.2 Attention Mining 214 2.3 Dual Consistency Loss 214 3 Experiments 215 3.1 Dataset 215 3.2 Experimental Settings 216 3.3 Experimental Results 216 4 Conclusion 219 References 219 Author Index 221 Front Matter ....Pages i-xii Context-Aware Synergetic Multiplex Network for Multi-organ Segmentation of Cervical Cancer MRI (Nesrine Bnouni, Islem Rekik, Mohamed Salah Rhim, Najoua Essoukri Ben Amara)....Pages 1-11 Residual Embedding Similarity-Based Network Selection for Predicting Brain Network Evolution Trajectory from a Single Observation (Ahmet Serkan Göktaş, Alaa Bessadok, Islem Rekik)....Pages 12-23 Adversarial Brain Multiplex Prediction from a Single Network for High-Order Connectional Gender-Specific Brain Mapping (Ahmed Nebli, Islem Rekik)....Pages 24-34 Learned Deep Radiomics for Survival Analysis with Attention (Ludivine Morvan, Cristina Nanni, Anne-Victoire Michaud, Bastien Jamet, Clément Bailly, Caroline Bodet-Milin et al.)....Pages 35-45 Robustification of Segmentation Models Against Adversarial Perturbations in Medical Imaging (Hanwool Park, Amirhossein Bayat, Mohammad Sabokrou, Jan S. Kirschke, Bjoern H. Menze)....Pages 46-57 Joint Clinical Data and CT Image Based Prognosis: A Case Study on Postoperative Pulmonary Venous Obstruction Prediction (Xinrong Hu, Zeyang Yao, Furong Liu, Wen Xie, Hailong Qiu, Haoyu Dong et al.)....Pages 58-67 Low-Dose CT Denoising Using Octave Convolution with High and Low Frequency Bands (Dong Kyu Won, Sion An, Sang Hyun Park, Dong Hye Ye)....Pages 68-78 Conditional Generative Adversarial Network for Predicting 3D Medical Images Affected by Alzheimer’s Diseases (Euijin Jung, Miguel Luna, Sang Hyun Park)....Pages 79-90 Inpainting Cropped Diffusion MRI Using Deep Generative Models (Rafi Ayub, Qingyu Zhao, M. J. Meloy, Edith V. Sullivan, Adolf Pfefferbaum, Ehsan Adeli et al.)....Pages 91-100 Multi-view Brain HyperConnectome AutoEncoder for Brain State Classification (Alin Banka, Inis Buzi, Islem Rekik)....Pages 101-110 Foreseeing Brain Graph Evolution over Time Using Deep Adversarial Network Normalizer (Zeynep Gürler, Ahmed Nebli, Islem Rekik)....Pages 111-122 Longitudinal Prediction of Radiation-Induced Anatomical Changes of Parotid Glands During Radiotherapy Using Deep Learning (Donghoon Lee, Sadegh Alam, Saad Nadeem, Jue Jiang, Pengpeng Zhang, Yu-Chi Hu)....Pages 123-132 Deep Parametric Mixtures for Modeling the Functional Connectome (Nicolas Honnorat, Adolf Pfefferbaum, Edith V. Sullivan, Kilian M. Pohl)....Pages 133-143 Deep EvoGraphNet Architecture for Time-Dependent Brain Graph Data Synthesis from a Single Timepoint (Ahmed Nebli, Uğur Ali Kaplan, Islem Rekik)....Pages 144-155 Uniformizing Techniques to Process CT Scans with 3D CNNs for Tuberculosis Prediction (Hasib Zunair, Aimon Rahman, Nabeel Mohammed, Joseph Paul Cohen)....Pages 156-168 mr\(^2\)NST: Multi-resolution and Multi-reference Neural Style Transfer for Mammography (Sheng Wang, Jiayu Huo, Xi Ouyang, Jifei Che, Zhong Xue, Dinggang Shen et al.)....Pages 169-177 Template-Oriented Multi-task Sparse Low-Rank Learning for Parkinson’s Diseases Diagnosis (Zihao Chen, Haijun Lei, Yujia Zhao, Zhongwei Huang, Xiaohua Xiao, Yi Lei et al.)....Pages 178-187 Multimodal Prediction of Breast Cancer Relapse Prior to Neoadjuvant Chemotherapy Treatment (Simona Rabinovici-Cohen, Ami Abutbul, Xosé M. Fernández, Oliver Hijano Cubelos, Shaked Perek, Tal Tlusty)....Pages 188-199 A Self-ensembling Framework for Semi-supervised Knee Cartilage Defects Assessment with Dual-Consistency (Jiayu Huo, Liping Si, Xi Ouyang, Kai Xuan, Weiwu Yao, Zhong Xue et al.)....Pages 200-209 Back Matter ....Pages 211-212 This book constitutes the proceedings of the Third International Workshop on Predictive Intelligence in Medicine, PRIME 2020, held in conjunction with MICCAI 2020, in Lima, Peru, in October 2020. The workshop was held virtually due to the COVID-19 pandemic. The 17 full and 2 short papers presented in this volume were carefully reviewed and selected for inclusion in this book. The contributions describe new cutting-edge predictive models and methods that solve challenging problems in the medical field for a high-precision predictive medicine.