Simulation and Synthesis in Medical Imaging: 7th International Workshop, SASHIMI 2022, Held in Conjunction with MICCAI 2022, Singapore, September 18, ... (Lecture Notes in Computer Science)

This book constitutes the refereed proceedings of the 7th International Workshop on Simulation and Synthesis in Medical Imaging, SASHIMI 2022, held in conjunction with MICCAI 2022, in Singapore, Singapore in September 2022. Preface Organization Contents Subject-Specific Lesion Generation and Pseudo-Healthy Synthesis for Multiple Sclerosis Brain Images 1 Introduction 1.1 Related Works 1.2 Contributions 2 Methods 2.1 Generators 2.2 Discriminators 2.3 Losses 3 Experiments 3.1 Evaluation 3.2 Implementation 3.3 Data 3.4 Results 4 Conclusion References Generating Artificial Artifacts for Motion Artifact Detection in Chest CT 1 Introduction 2 Methods 3 Experiments 4 Results 5 Discussion References Probabilistic Image Diversification to Improve Segmentation in 3D Microscopy Image Data 1 Introduction 2 Probabilistic Image Diversification 3 Experiments and Results 3.1 Data Augmentation 3.2 Benchmarking 3.3 Test-Time Augmentation 4 Discussion and Conclusion References Pathology Synthesis of 3D Consistent Cardiac MR Images Using 2D VAEs and GANs 1 Introduction 1.1 Contributions 2 Method 2.1 Pathology Synthesis 2.2 Modeling Slice Relationship 2.3 Data and Implementation 3 Results 3.1 Pathology Synthesis 3.2 Modeling the Slice Relationship 4 Discussion and Conclusion References .26em plus .1em minus .1emHealthyGAN: Learning from Unannotated Medical Images to Detect Anomalies Associated with Human Disease 1 Introduction 2 HealthyGAN: The Proposed Method 2.1 Network Architecture 2.2 Training 2.3 Detecting Anomalies 3 Experiments and Results 3.1 COVID-19 Detection 3.2 Chest X-ray 14 Diseases Detection 3.3 Migraine Detection 4 Conclusion A Implementation Details B Network Architectures B.1 Discriminator B.2 Generator References Bi-directional Synthesis of Pre- and Post-contrast MRI via Guided Feature Disentanglement 1 Introduction 2 Methodology 3 Experiments and Results 4 Conclusion References Morphology-Preserving Autoregressive 3D Generative Modelling of the Brain 1 Introduction 2 Background 2.1 VQ-VAE 2.2 Transformer 3 Methods 3.1 Descriptive Quantization for Transformer Usage 3.2 Autoregressive Modelling of the Brain 4 Experiments and Results 4.1 Quantitative Image Fidelity Evaluation 4.2 Morphological Evaluation 5 Conclusion 6 Appendix 6.1 VQ-VAEs 6.2 Transformers 6.3 Losses 6.4 Datasets 6.5 VBM Analysis References Can Segmentation Models Be Trained with Fully Synthetically Generated Data? 1 Background 2 Materials and Methods 2.1 Materials 2.2 Methods 2.3 Segmentation Network Used for the Experiments 3 Experiments 3.1 Can We Learn to Segment Healthy Regions Using Synthetic Data? 3.2 Can Synthetic Generative Models Address Out-of-Distribution Segmentation? 3.3 Can We Learn to Segment Pathologies from Synthetic Data? 4 Discussion and Conclusion A Training Set-Ups A.1 Training brainSPADE A.2 Training Segmentation nnU-Nets B Additional Figures References Multimodal Super Resolution with Dual Domain Loss and Gradient Guidance 1 Introduction 2 Materials and Methods 3 Experiments 4 Discussion and Conclusion References Brain Lesion Synthesis via Progressive Adversarial Variational Auto-Encoder 1 Introduction 2 Methodology 2.1 Model Architecture 2.2 Condition and Mask Embedding Blocks 2.3 Loss Functions 3 Experiments 3.1 Dataset 3.2 Implementation Details 3.3 Evaluation Metrics 3.4 Experimental Results 4 Discussion and Conclusion References Contrastive Learning for Generating Optical Coherence Tomography Images of the Retina 1 Introduction 2 Related Work 3 Methods 4 Experiments and Results 4.1 Dataset 4.2 Model Training 4.3 Results 5 Conclusions References A Novel Method Combining Global and Local Assessments to Evaluate CBCT-Based Synthetic CTs 1 Introduction 2 Methods 2.1 Data Acquisition and Processing 2.2 Validation Methodology 3 Results 3.1 Validation Methodology 4 Discussion 5 Conclusion References SuperFormer: Volumetric Transformer Architectures for MRI Super-Resolution 1 Introduction 2 Method 2.1 Feature Embedding 2.2 Volume Embedding 2.3 3D Deep Feature Extraction 2.4 HQ Volume Reconstruction 3 Experimental Setup 3.1 Implementation Details 3.2 Results 4 Conclusion References Evaluating the Performance of StyleGAN2-ADA on Medical Images 1 Introduction 2 Methods 2.1 Data 2.2 Generative Modeling 2.3 Evaluation Measures 3 Results 4 Conclusion References Backdoor Attack is a Devil in Federated GAN-Based Medical Image Synthesis 1 Introduction 2 Methods 2.1 Federated Generative Adversarial Network 2.2 Backdoor Attack Strategies 2.3 Defense Strategies 3 Experiments 3.1 Experimental Settings 3.2 Implementation of Attack 3.3 Implementation of Defense 3.4 Results and Discussion 4 Conclusion A More Experiment Results B WGAN-GP with Large Trigger Size References Author Index