Myocardial Pathology Segmentation Combining Multi-Sequence Cardiac Magnetic Resonance Images : First Challenge, MyoPS 2020, Held in Conjunction with MICCAI 2020, Lima, Peru, October 4, 2020, Proceedings

This book constitutes the First Myocardial Pathology Segmentation Combining Multi-Sequence CMR Challenge, MyoPS 2020, which was held in conjunction with the 23rd International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2020, in Lima, Peru, in October 2020. The challenge took place virtually due to the COVID-19 crisis. The 12 full and 4 short papers presented in this volume were carefully reviewed and selected form numerous submissions. This challenge aims not only to benchmark various myocardial pathology segmentation algorithms, but also to cover the topic of general cardiac image segmentation, registration and modeling, and raise discussions for further technical development and clinical deployment. Preface Organization Contents Stacked BCDU-Net with Semantic CMR Synthesis: Application to Myocardial Pathology Segmentation Challenge 1 Introduction 2 Materials and Methods 2.1 Dataset 2.2 Proposed Method 2.3 Data Augmentation Strategy 2.4 Post-processing 3 Results 3.1 Protocol and Metrics of the Challenge 3.2 Ablation Study 3.3 Challenge Results 4 Discussion References EfficientSeg: A Simple But Efficient Solution to Myocardial Pathology Segmentation Challenge 1 Introduction 2 Dataset 3 Method 3.1 Encoder 3.2 Decoder 3.3 Optimization 3.4 Emplementation Details 4 Results 5 Conclusion References Two-Stage Method for Segmentation of the Myocardial Scars and Edema on Multi-sequence Cardiac Magnetic Resonance 1 Introduction 2 Method 2.1 Dataset Description 2.2 Image Segmentation 3 Experiments and Results 3.1 Experimental Configuration 3.2 Performance Evaluation and Analysis 4 Conclusion References Multi-modality Pathology Segmentation Framework: Application to Cardiac Magnetic Resonance Images 1 Introduction 2 Method 2.1 Anatomical Structure Segmentation Network (ASSN) 2.2 Pathological Region Segmentation Network (PRSN) 3 Experiment 3.1 Dataset 3.2 Implementations 3.3 Results 4 Conclusion References Myocardial Edema and Scar Segmentation Using a Coarse-to-Fine Framework with Weighted Ensemble 1 Introduction 2 Method 2.1 Data Preprocessing 2.2 Coarse Segmentation Network 2.3 Fine Segmentation Network 2.4 Weighted Ensemble 3 Experiments and Results 3.1 Data and Implementation 3.2 Results 4 Discussion and Conclusion References Exploring Ensemble Applications for Multi-sequence Myocardial Pathology Segmentation 1 Introduction 1.1 Background 1.2 Related Work 2 Experiments 3 Methods 3.1 Software 3.2 Processing Pipeline and Architecture 3.3 Hyper-parameter Search 3.4 Ensemble Method 4 Results 4.1 Cross-Validation Results on Training Set 4.2 Performance on Test Set 5 Discussion References Max-Fusion U-Net for Multi-modal Pathology Segmentation with Attention and Dynamic Resampling 1 Introduction 2 Methodology 2.1 Model Architecture 3 Implementation 3.1 Dynamic Resampling Training Strategy 3.2 Training with Alternative Cross Validation 4 Experiments 4.1 Results and Discussion 4.2 Prediction for the Challenge Testing Dataset 5 Conclusions References Fully Automated Deep Learning Based Segmentation of Normal, Infarcted and Edema Regions from Multiple Cardiac MRI Sequences 1 Introduction 2 Methodology 2.1 Augmentation Module 2.2 Preprocessing 2.3 Linear Encoder 2.4 Network Module 2.5 Post-processing 2.6 Linear Decoder 3 Experiments 3.1 Clinical Data 3.2 Implementation Details 3.3 Evaluation Metrics 4 Results 4.1 Quantitative Assessment 4.2 Visual Assessment 5 Conclusion References CMS-UNet: Cardiac Multi-task Segmentation in MRI with a U-Shaped Network 1 Introduction 2 Method 2.1 Shared Encoder 2.2 Channel Reconstruction Upsampling 2.3 Multi-scale Convolution Module 2.4 Loss Function 3 Experimental Results 3.1 Dataset 3.2 Implementation Details 3.3 Results 3.4 Ablation Study 4 Conclusion References Automatic Myocardial Scar Segmentation from Multi-sequence Cardiac MRI Using Fully Convolutional Densenet with Inception and Squeeze-Excitation Module 1 Introduction 2 Materials 3 Methods 3.1 Proposed Pipeline 3.2 Network Architecture 3.3 Region of Interest (ROI) Detection 3.4 Myocardium and Left Ventricle Segmentation 3.5 Scar Segmentation 3.6 Loss Function 3.7 Training 4 Results and Discussion 4.1 Myocardium and Left Ventricle 4.2 Scar 5 Conclusion References Dual Attention U-Net for Multi-sequence Cardiac MR Images Segmentation 1 Introduction 2 Method 2.1 Data Processing 2.2 Dual Attention U-Net Architecture 2.3 Post Processing 3 Experiments and Results 3.1 Data and Evaluation Metrics 3.2 Implementation Details 3.3 Experimental Results and Analysis 4 Conclusion References Accurate Myocardial Pathology Segmentation with Residual U-Net 1 Introduction 2 Background 3 Methods 3.1 Dataset 3.2 U-Net Architecture 3.3 Residual U-Net Architecture 3.4 Loss Function and Evaluation Metric 3.5 Implementation Details 4 Experimental Results 4.1 K-Fold Cross-validation Results 4.2 Test Results 5 Discussion and Conclusion References Stacked and Parallel U-Nets with Multi-output for Myocardial Pathology Segmentation 1 Introduction 2 Methodology 2.1 Overview of Network Architecture 3 Experimental Results 3.1 Segmentation Results 4 Conclusion References Dual-Path Feature Aggregation Network Combined Multi-layer Fusion for Myocardial Pathology Segmentation with Multi-sequence Cardiac MR 1 Introduction 2 Methodology 2.1 Data Processing 2.2 Proposed Method 3 Experiments and Results 3.1 Implementation Details 3.2 Ablation Experiments and Results 3.3 Test Results 4 Conclusion References Cascaded Framework with Complementary CMR Information for Myocardial Pathology Segmentation 1 Introduction 2 Method 3 Experiments and Results 3.1 Dataset and Training Protocols 3.2 Five-Fold Cross Validation Results of the Whole LV Segmentation 3.3 Five-Fold Cross Validation Results of the Pathology Segmentation 3.4 Pathology Segmentation Results on Testing Set 4 Conclusion References Recognition and Standardization of Cardiac MRI Orientation via Multi-tasking Learning and Deep Neural Networks 1 Introduction 2 Method 3 Experiment 4 Conclusion References Correction to: Two-Stage Method for Segmentation of the Myocardial Scars and Edema on Multi-sequence Cardiac Magnetic Resonance Correction to: Chapter “Two-Stage Method for Segmentation of the Myocardial Scars and Edema on Multi-sequence Cardiac Magnetic Resonance” in: X. Zhuang and L. Li (Eds.): Myocardial Pathology Segmentation Combining Multi-Sequence Cardiac Magnetic Resonance Images, LNCS 12554, https://doi.org/10.1007/978-3-030-65651-5_3 Author Index