Medical Image Computing and Computer Assisted Intervention – MICCAI 2021 : 24th International Conference, Strasbourg, France, September 27 – October 1, 2021, Proceedings, Part V

The eight-volume set LNCS 12901, 12902, 12903, 12904, 12905, 12906, 12907, and 12908 constitutes the refereed proceedings of the 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021, held in Strasbourg, France, in September/October 2021.* The 531 revised full papers presented were carefully reviewed and selected from 1630 submissions in a double-blind review process. The papers are organized in the following topical sections: Part I: image segmentation Part II: machine learning - self-supervised learning; machine learning - semi-supervised learning; and machine learning - weakly supervised learning Part III: machine learning - advances in machine learning theory; machine learning - attention models; machine learning - domain adaptation; machine learning - federated learning; machine learning - interpretability / explainability; and machine learning - uncertainty Part IV: image registration; image-guided interventions and surgery; surgical data science; surgical planning and simulation; surgical skill and work flow analysis; and surgical visualization and mixed, augmented and virtual reality Part V: computer aided diagnosis; integration of imaging with non-imaging biomarkers; and outcome/disease prediction Part VI: image reconstruction; clinical applications - cardiac; and clinical applications - vascular Part VII: clinical applications - abdomen; clinical applications - breast; clinical applications - dermatology; clinical applications - fetal imaging; clinical applications - lung; clinical applications - neuroimaging - brain development; clinical applications - neuroimaging - DWI and tractography; clinical applications - neuroimaging - functional brain networks; clinical applications - neuroimaging – others; and clinical applications - oncology Part VIII: clinical applications - ophthalmology; computational (integrative) pathology; modalities - microscopy; modalities - histopathology; and modalities - ultrasound *The conference was held virtually. Preface Organization Contents – Part V Computer Aided Diagnosis DeepStationing: Thoracic Lymph Node Station Parsing in CT Scans Using Anatomical Context Encoding and Key Organ Auto-Search 1 Introduction 2 Method 2.1 Stratified Chest Organ Segmentation 2.2 Anatomical Context Encoded LNS Parsing 3 Experimental Results 4 Conclusion References Hepatocellular Carcinoma Segmentation from Digital Subtraction Angiography Videos Using Learnable Temporal Difference 1 Introduction 2 Methodology 2.1 Key Frame Selection and Segmentation 2.2 Temporal Difference Learning 2.3 Liver Region Guidance 2.4 Architecture and Loss Function 3 Dataset Construction 4 Experiments and Results 4.1 Implementation Details 4.2 Quantitative Analysis 4.3 Qualitative Results 5 Discussions and Conclusions References CA-Net: Leveraging Contextual Features for Lung Cancer Prediction 1 Introduction 1.1 Related Work 2 Methodology 2.1 Nodule Detection 2.2 Nodule Malignancy Classification 2.3 Cancer Prediction 3 Experiments 3.1 Experimental Results 3.2 Ablation Study 3.3 Visualization 4 Conclusions References Semi-supervised Learning for Bone Mineral Density Estimation in Hip X-Ray Images 1 Introduction 2 Method 2.1 Adaptive Triplet Loss 2.2 Semi-supervised Self-training 3 Experiment Results 4 Conclusion References DAE-GCN: Identifying Disease-Related Features for Disease Prediction 1 Introduction 2 Methodology 2.1 Encoder 2.2 Disentangle Training 3 Experiments 3.1 Results 3.2 Ablation Study 3.3 Interpretability 4 Conclusions and Disscusions References Enhanced Breast Lesion Classification via Knowledge Guided Cross-Modal and Semantic Data Augmentation 1 Introduction 2 Methodology 2.1 Baseline Classification Networks 2.2 Modal Translater 2.3 Semantic Inverter 2.4 Final Classification Network 3 Experiments 3.1 Dataset Preparation 3.2 Implementation Details 3.3 Experimental Results 3.4 Ablation Study 4 Conclusion References Multiple Meta-model Quantifying for Medical Visual Question Answering 1 Introduction 2 Literature Review 3 Methodology 3.1 Method Overview 3.2 Multiple Meta-model Quantifying 3.3 Integrate Quantified Meta-models to Medical VQA Framework 4 Experiments 4.1 Dataset 4.2 Experimental Details 4.3 Results 4.4 Ablation Study 5 Conclusion References mfTrans-Net: Quantitative Measurement of Hepatocellular Carcinoma via Multi-Function Transformer Regression Network 1 Introduction 2 Method 2.1 CNN-based Encoder for Feature Extraction and Dimension Reducing 2.2 mf-Trans for Self-attention and Phase-Aware 2.3 Multi-level Constraint Strategy of Enhanced Loss Function 3 Experiment and Results 3.1 Multi-index Quantification Comparison 3.2 Performance in Different Size of HCC 3.3 Ablation Study 4 Conclusion References You only Learn Once: Universal Anatomical Landmark Detection 1 Introduction 2 Method 2.1 The Local Network 2.2 The Global Network 2.3 Loss Function 3 Experiments 3.1 Settings 3.2 Dataset 3.3 Evaluation 3.4 Ablation Study 4 Conclusions References A Coherent Cooperative Learning Framework Based on Transfer Learning for Unsupervised Cross-Domain Classification 1 Introduction 2 Method 2.1 Wasserstein CycleGAN in the Proposed Method 2.2 Cooperative Training of Classifiers Based on Transfer Learning 2.3 Prediction 3 Experiments 3.1 Databases and Settings 3.2 Performance of the Supervised Classifier 3.3 Evaluation of Unsupervised Cross-Domain Classification 3.4 Visualization and Ablation Experiments 4 Discussion and Conclusion References Towards a Non-invasive Diagnosis of Portal Hypertension Based on an Eulerian CFD Model with Diffuse Boundary Conditions 1 Introduction 2 Methodology 2.1 Blood Flow Model 2.2 Diffuse Boundary Conditions 2.3 Data Pre-processing and Numerical Implementation 3 Experiments and Results 3.1 In Silico Study on a Steady Poiseuille Flow 3.2 In-Vivo Validation with Invasive HVPG 4 Conclusion References A Segmentation-Assisted Model for Universal Lesion Detection with Partial Labels 1 Introduction 2 Method 2.1 Joint Learning with Semantic Segmentation 2.2 Fine-Tuning with Mined Lesions 3 Experiments 3.1 Experiment Setup 3.2 Comparison with State-of-the-Arts 3.3 Analysis of Our Method 4 Conclusion References Constrained Contrastive Distribution Learning for Unsupervised Anomaly Detection and Localisation in Medical Images 1 Introduction 2 Method 2.1 Constrained Contrastive Distribution Learning 2.2 Anomaly Detection and Localisation 3 Experiments 3.1 Dataset 3.2 Implementation Details 3.3 Ablation Study 3.4 Comparison to SOTA Models 4 Conclusion References Conditional Training with Bounding Map for Universal Lesion Detection 1 Introduction 2 Method 2.1 Bounding Map Generation 2.2 BM-based Conditioning Mechanism 2.3 Pseudo Lesion Segmentation via ABM Supervision 3 Experiments 3.1 Dataset and Setting 3.2 Lesion Detection Performance 3.3 Ablation Study 4 Conclusion References Focusing on Clinically Interpretable Features: Selective Attention Regularization for Liver Biopsy Image Classification 1 Introduction 2 Proposed Method 2.1 Image Classification 2.2 Selective Attention Regularization 3 Experiments 3.1 Experiment Settings 3.2 Quantitative Analysis 3.3 Qualitative Analysis 4 Conclusion References Categorical Relation-Preserving Contrastive Knowledge Distillation for Medical Image Classification 1 Introduction 2 Method 2.1 Class-Guided Contrastive Distillation (CCD) 2.2 Categorical Relation Preserving (CRP) 2.3 Training and Testing 3 Experiments 3.1 Dataset and Implementation Details 3.2 Experimental Results 4 Conclusion References Tensor-Based Multi-index Representation Learning for Major Depression Disorder Detection with Resting-State fMRI 1 Introduction 2 Materials and Method 2.1 Subjects and Image Pre-processing 2.2 Methodology 3 Experiment and Results 4 Conclusion References Region Ensemble Network for MCI Conversion Prediction with a Relation Regularized Loss 1 Introduction 2 Method 2.1 Region Ensemble Network 2.2 Relation Regularized Loss 3 Experiments 3.1 Dataset and Evaluation Metrics 3.2 Implementation Details 3.3 Ablation Studies 3.4 Comparing with SOTA Methods 3.5 Interpreting the Model's Prediction 4 Conclusion References Airway Anomaly Detection by Prototype-Based Graph Neural Network 1 Introduction 2 Related Work 2.1 Airway-Related CAD System 2.2 Anomaly Detection 2.3 Graph Neural Network 3 Method 3.1 Structured Feature Extraction 3.2 Graph Neural Network 3.3 Prototype-Based Anomaly Detection 4 Experiments 5 Conclusion References Energy-Based Supervised Hashing for Multimorbidity Image Retrieval 1 Introduction 2 Methodology 2.1 Network Architecture 2.2 Learning Objectives 2.3 Training Algorithm 3 Experiments and Results 3.1 Dataset 3.2 Experiment Settings 3.3 Results and Analysis 4 Conclusion References Stochastic 4D Flow Vector-Field Signatures: A New Approach for Comprehensive 4D Flow MRI Quantification 1 Introduction 2 Methodology 2.1 Stage I: Stochastic 4D Flow Vector-Field Signature Construction 2.2 Stage II: Hemodynamic Signature Index (HSI) 3 Datasets and Preprocessing 4 Convergence Analysis of Optimal Sampling Density Factor (α) 5 Signature Sensitivity Analysis to Segmentation Errors 6 In Vivo Scan-Rescan Reproducibility Study 7 Signature Comparison of BAV Patients and Healthy Controls 8 Results 9 Discussion and Conclusions References Source-Free Domain Adaptive Fundus Image Segmentation with Denoised Pseudo-Labeling 1 Introduction 2 Method 2.1 Pseudo-Labeling for Source-Free UDA 2.2 Pixel-Level Denoising via Uncertainty Estimation 2.3 Class-Level Denoising via Prototype Estimation 2.4 Training Procedures and Implementation Details 3 Experiments 4 Conclusion References ASC-Net: Adversarial-Based Selective Network for Unsupervised Anomaly Segmentation 1 Introduction 2 Adversarial-Based Selective Cutting Network (ASC-Net) 2.1 Network Framework 2.2 Architecture Details and Training Scheme 3 Applications 4 Discussion and Future Work References Cost-Sensitive Meta-learning for Progress Prediction of Subjective Cognitive Decline with Brain Structural MRI 1 Introduction 2 Materials and Methodology 3 Experiment 4 Conclusion References Effective Pancreatic Cancer Screening on Non-contrast CT Scans via Anatomy-Aware Transformers 1 Introduction 1.1 Related Work 2 Methodology 2.1 Anatomy-Aware Classification with Transformers 3 Experiments 4 Conclusion References Learning from Subjective Ratings Using Auto-Decoded Deep Latent Embeddings 1 Introduction 2 Methods 3 Experiments 4 Conclusion References VertNet: Accurate Vertebra Localization and Identification Network from CT Images 1 Introduction 2 Method 2.1 Vertebrae Localization 2.2 Vertebrae Identification 2.3 Implementation Details 3 Experiments 3.1 Dataset and Evaluation Metric 3.2 Ablation Study of Key Components 3.3 Comparison with State-of-the-Art Methods 4 Conclusion References VinDr-SpineXR: A Deep Learning Framework for Spinal Lesions Detection and Classification from Radiographs 1 Introduction 1.1 Spine X-Ray Interpretation 1.2 Related Work 1.3 Contribution 2 Proposed Method 2.1 Overall Framework 2.2 Dataset 2.3 Model Development 3 Experiments and Results 3.1 Experimental Setup and Implementation Details 3.2 Evaluation Metrics 3.3 Experimental Results 4 Discussion 5 Conclusions References Multi-frame Collaboration for Effective Endoscopic Video Polyp Detection via Spatial-Temporal Feature Transformation 1 Introduction 2 Method 3 Experiments 3.1 Datasets and Settings 3.2 Quantitative and Qualitative Comparison 3.3 Ablation Study 4 Conclusion References MBFF-Net: Multi-Branch Feature Fusion Network for Carotid Plaque Segmentation in Ultrasound 1 Introduction 2 Method 2.1 CAW Detecting Module 2.2 MBFF Module 2.3 Boundary Preserving Structure 3 Experiments 4 Conclusion References Balanced-MixUp for Highly Imbalanced Medical Image Classification 1 Introduction 2 Methodology 2.1 MixUp Regularization 2.2 Training Data Sampling Strategies 2.3 Balanced-MixUp 2.4 Training Details 3 Experimental Analysis 3.1 Experimental Details 3.2 Discussion on the Numerical Results on DR Grading 3.3 Discussion on the Numerical Results on GI Image Classification 4 Conclusion References Transfer Learning of Deep Spatiotemporal Networks to Model Arbitrarily Long Videos of Seizures-10pt 1 Introduction 2 Materials and Methods 2.1 Video Acquisition 2.2 Dataset Description and Ground-Truth Definitions 2.3 Snippet-Level Classification 2.4 Seizure-Level Classification 3 Experiments and Results 3.1 Evaluation of Feature Extractors for Snippets 3.2 Aggregation for Seizure-Level Classification 4 Discussion and Conclusion References Retina-Match: Ipsilateral Mammography Lesion Matching in a Single Shot Detection Pipeline 1 Introduction 2 Related Work 3 Method 4 Experiments 4.1 Dataset 4.2 Network Architecture 4.3 Training 4.4 Choice of Single-View Detection Model 4.5 Ablation Study 4.6 Comparison with the Relation Block Approach 5 Conclusion References Towards Robust Dual-View Transformation via Densifying Sparse Supervision for Mammography Lesion Matching 1 Introduction 2 Method 2.1 Spatial Location Determination in Fake Lesions 2.2 LT-GAN for Cross-View Lesion Synthesis 3 Experiment 3.1 Dataset and Implementation Details 3.2 Ablation Study 3.3 Comparison with Other DVT Learning Methods 4 Conclusions References DeepOPG: Improving Orthopantomogram Finding Summarization with Weak Supervision 1 Introduction 2 Methods 2.1 Model Architecture 2.2 Improved Tooth Localization with Weakly Supervised Reinforcement Learning 3 Experiments 3.1 Dataset 3.2 Overall Evaluation of DeepOPG for Findings Summarization 3.3 Tooth Localization with Dental Coherence 3.4 Comparing Existing Works 4 Conclusion References Joint Spinal Centerline Extraction and Curvature Estimation with Row-Wise Classification and Curve Graph Network 1 Introduction 2 Method 2.1 Spine Centerline Extraction by Row-Wise Classification 2.2 Curve Feature Pooling 2.3 Curve Graph Network (CGN) 3 Experiments 3.1 Dataset and Evaluation 3.2 Implementation Details 3.3 Result and Analysis 4 Conclusion References LDPolypVideo Benchmark: A Large-Scale Colonoscopy Video Dataset of Diverse Polyps-8pt 1 Introduction 2 Our LDPolypVideo Dataset 2.1 Data Acquisition 2.2 Tracking-Assisted Annotation Tool 3 Experiments on LDPolypVideo Dataset 4 Conclusion References Continual Learning with Bayesian Model Based on a Fixed Pre-trained Feature Extractor 1 Introduction 2 A Generative Model for Continual Learning 2.1 Fixed Pre-trained Feature Extractor 2.2 Memory Formation 2.3 Bayesian Model for Prediction 3 Experimental Evaluations 3.1 Experimental Setup 3.2 Effectiveness of the Generative Model 3.3 Generalizability and Robustness of the Generative Model 3.4 Effect of Feature Extractor 4 Conclusion References Alleviating Data Imbalance Issue with Perturbed Input During Inference 1 Introduction 2 Methodology 3 Experiments 3.1 Experimental Settings 3.2 Effectiveness and Generalizability Evaluation 3.3 Robustness to Hyper-parameters 4 Conclusion References A Deep Reinforced Tree-Traversal Agent for Coronary Artery Centerline Extraction 1 Introduction 2 Methods 3 Experiments and Results 4 Conclusion and Future Work References Sequential Gaussian Process Regression for Simultaneous Pathology Detection and Shape Reconstruction 1 Introduction 1.1 Related Work 1.2 Main Contributions 2 Background 3 Method: Sequential GPMM Regression 4 Evaluation 5 Conclusion References Predicting Symptoms from Multiphasic MRI via Multi-instance Attention Learning for Hepatocellular Carcinoma Grading 1 Introduction 2 Method 2.1 Model Overview 2.2 Modules 2.3 Implementation Details 3 Experiment 3.1 Dataset and Experimental Setup 3.2 Results 4 Conclusion References Triplet-Branch Network with Prior-Knowledge Embedding for Fatigue Fracture Grading 1 Introduction 2 Method 2.1 Cumulative Learning Module 2.2 Auxiliary Ranking Task 3 Experiments 3.1 Ablation Study 3.2 Comparison with State-of-the-art 3.3 Generalization Evaluation 4 Conclusion References DeepMitral: Fully Automatic 3D Echocardiography Segmentation for Patient Specific Mitral Valve Modelling 1 Introduction 2 Methods 2.1 Data Acquisition 2.2 Model Selection 2.3 DeepMitral Pipeline 2.4 Evaluation 3 Results 3.1 Inference Runtime Performance 4 Discussion 4.1 Limitations and Future Work 5 Conclusions References Data Augmentation in Logit Space for Medical Image Classification with Limited Training Data 1 Introduction 2 Methodology 2.1 Classifier with Logit Uncertainty for Data Augmentation 2.2 Comparison with Relevant Techniques 3 Experiment 3.1 Experimental Setting 3.2 Effectiveness Evaluation 3.3 Model Component Choice and Effect of Hyper-parameters 4 Conclusion References Collaborative Image Synthesis and Disease Diagnosis for Classification of Neurodegenerative Disorders with Incomplete Multi-modal Neuroimages 1 Introduction 2 Method 2.1 Problem Formulation 2.2 Model Overview 2.3 Image Synthesis 2.4 Multi-modal Diagnosis 2.5 Collaborative Learning 3 Materials 4 Experiments and Results 4.1 Experimental Settings 4.2 Results of Neuroimage Synthesis 4.3 Results of Disease Diagnosis 5 Conclusion References Seg4Reg+: Consistency Learning Between Spine Segmentation and Cobb Angle Regression 1 Introduction 2 Method 3 Experiments 4 Conclusion References Meta-modulation Network for Domain Generalization in Multi-site fMRI Classification 1 Introduction 2 Materials and Image Preprocessing 3 Proposed Method 3.1 Problem Setup 3.2 Overview Method 3.3 Meta-Learning Modulation Network 3.4 Training the Domain Generalized Model 4 Experiments and Analysis 4.1 Experimental Settings 4.2 Results and Discussion 5 Conclusion References 3D Brain Midline Delineation for Hematoma Patients 1 Introduction 2 Methodology 2.1 Hemisphere Segmentation Network 2.2 Distance-Weighted Map 2.3 Rectificative Learning 2.4 Midline Correction 3 Experiments 3.1 Dataset and Implementation Details 3.2 Quantitative and Qualitative Evaluation 4 Conclusion References Unsupervised Representation Learning Meets Pseudo-Label Supervised Self-Distillation: A New Approach to Rare Disease Classification 1 Introduction 2 Methods 3 Experiments 4 Conclusion References nnDetection: A Self-configuring Method for Medical Object Detection 1 Introduction 2 Methods 3 Experiments and Results 4 Discussion References Automating Embryo Development Stage Detection in Time-Lapse Imaging with Synergic Loss and Temporal Learning 1 Introduction 2 Related Work 3 Methodology 3.1 Synergic Network 3.2 Temporal Learning 3.3 Pre- and Post-processing 4 Experiments 4.1 Dataset Overview 4.2 Implementation Details 4.3 Experimental Results 5 Discussion 6 Conclusions References Deep Neural Dynamic Bayesian Networks Applied to EEG Sleep Spindles Modeling 1 Introduction 2 Generative Model for Single–channel EEG 2.1 Learning Model Parameters 3 Results 3.1 Validation with Access to Ground Truth 3.2 Sleep Spindles Density Varies Throughout Sleep Stages 4 Conclusion References Few Trust Data Guided Annotation Refinement for Upper Gastrointestinal Anatomy Recognition 1 Introduction 2 Methods 2.1 Adaptive Weighted Loss 2.2 Trust Data Guided Annotation Refinement Framework 3 Experiments 3.1 Controlled Experiments 3.2 Experiments on Upper GI Dataset 4 Conclusion References Asymmetric 3D Context Fusion for Universal Lesion Detection 1 Introduction 2 Methods 2.1 Preliminary: 3D Context Fusion Operators with 2D Pretraining 2.2 Asymmetric 3D Context Fusion (A3D) 2.3 Network Structure for Universal Lesion Detection 3 Experiments 3.1 Dataset and Experiment Settings 3.2 Performance Analysis 4 Conclusion References Detecting Outliers with Poisson Image Interpolation 1 Introduction 2 Method 3 Evaluation and Results 4 Conclusion References MG-NET: Leveraging Pseudo-imaging for Multi-modal Metagenome Analysis 1 Introduction 2 MG-NET: Leveraging Pseudo-imaging for Multimodal Metagenome Analysis 2.1 Capturing the Global Structure with Graph Representations 2.2 Pseudo-Imaging for Local Structural Properties 2.3 Structural Reasoning with Attention 2.4 Implementation Details 3 Experimental Evaluation 3.1 Quantitative Evaluation 4 Conclusion and Future Work References Multimodal Multitask Deep Learning for X-Ray Image Retrieval 1 Introduction 2 Related Work 3 Methods 4 Experiment Setup 5 Results 6 Conclusion References Linear Prediction Residual for Efficient Diagnosis of Parkinson's Disease from Gait 1 Introduction 2 Materials and Methods 2.1 Dataset 2.2 Data-Leakage Experiment 2.3 Model Pipeline 2.4 Evaluation 3 Results and Discussion 4 Conclusion References Primary Tumor and Inter-Organ Augmentations for Supervised Lymph Node Colon Adenocarcinoma Metastasis Detection 1 Introduction 2 Related Work 3 Method 4 Results 4.1 Primary Tumor Is an Inexpensive Training Data Source for Lymph Node Colon Cancer Metastasis Detection 4.2 Domain Adapted Data Closes the Gap for Large Representation Shift Between Source and Target Domain 4.3 Robustness Is Improved by Out-of-Domain Data 5 Conclusion References Radiomics-Informed Deep Curriculum Learning for Breast Cancer Diagnosis 1 Introduction 2 Methods 3 Evaluation 3.1 Evaluation Settings of the Proposed Method 3.2 Dataset, Experimental Setup, and Performance Metrics 4 Experiment Results 5 Conclusions References Integration of Imaging with Non-Imaging Biomarkers Lung Cancer Risk Estimation with Incomplete Data: A Joint Missing Imputation Perspective 1 Introduction 2 Theory 3 Experiment Designs and Results 4 Conclusion References Co-graph Attention Reasoning Based Imaging and Clinical Features Integration for Lymph Node Metastasis Prediction 1 Introduction 2 Dataset 3 Method 3.1 Graph Node Attributes Construction with Category-Wise Attention 3.2 Co-graph Attention Based Reasoning 4 Experiments 5 Conclusion References Deep Orthogonal Fusion: Multimodal Prognostic Biomarker Discovery Integrating Radiology, Pathology, Genomic, and Clinical Data 1 Introduction 2 Methodology 2.1 Loss Functions 2.2 Modality-Specific Networks for Outcome Prediction 3 Experimental Details 4 Results and Discussion 5 Conclusions References A Novel Bayesian Semi-parametric Model for Learning Heritable Imaging Traits 1 Introduction 2 Method 3 Experiments and Results 4 Conclusion References Combining 3D Image and Tabular Data via the Dynamic Affine Feature Map Transform 1 Introduction 2 Related Work 3 Methods 4 Experiments 4.1 Data Processing 4.2 Evaluation Scheme 5 Results 6 Conclusion References Image-Derived Phenotype Extraction for Genetic Discovery via Unsupervised Deep Learning in CMR Images 1 Introduction 1.1 Related Work 2 Methods 2.1 Description of the Data 2.2 Graph-Convolutional Autoencoder 2.3 GWAS 2.4 Proposed Algorithm 2.5 Implementation Details 3 Results and Discussion 4 Conclusions References GKD: Semi-supervised Graph Knowledge Distillation for Graph-Independent Inference 1 Introduction 2 Knowledge Distillation Framework 3 Methods 4 Experiments and Results 4.1 Autism Spectrum Disorder Diagnosis on the ABIDE Dataset 4.2 Alzheimer's Progression Prediction on the TADPOLE Dataset 4.3 Synthetic Dataset 4.4 Discussion and Conclusion References Outcome/Disease Prediction Predicting Esophageal Fistula Risks Using a Multimodal Self-attention Network 1 Introduction 2 Dataset 3 Multimodal Network with the VisText Self-attention Module 3.1 Multimodal Feature Extraction 3.2 VisText Self-attention Module for Multimodal Dependency Extraction 3.3 Aggregation and Optimization 4 Experiments and Results 4.1 Experimental Settings and Pre-processing 4.2 Comparison Methods and Evaluation Measures 4.3 Comparison with Others and Ablation Study Results 5 Conclusion References Hybrid Aggregation Network for Survival Analysis from Whole Slide Histopathological Images 1 Introduction 2 Method 2.1 Self-supervised Feature Extraction 2.2 Hybrid Aggregation Network 2.3 Loss 3 Experiment 3.1 Dataset 3.2 Implementation Details 3.3 Evaluation Metric 3.4 Ablation Study 3.5 Results 4 Conclusion References Intracerebral Haemorrhage Growth Prediction Based on Displacement Vector Field and Clinical Metadata 1 Introduction 2 Method 2.1 The DVF Based Framework 2.2 Fusing Clinical Metadata 2.3 Data Preprocessing and Augmentation 2.4 Implementation Details 3 Experiments and Results 4 Conclusion References AMINN: Autoencoder-Based Multiple Instance Neural Network Improves Outcome Prediction in Multifocal Liver Metastases 1 Introduction 2 Methods 3 Experiments 4 Results 5 Conclusion and Discussion References Survival Prediction Based on Histopathology Imaging and Clinical Data: A Novel, Whole Slide CNN Approach 1 Introduction 2 Methodology 2.1 Loss Function 2.2 Whole Slide Feature Maps (WSFM) 2.3 Siamese Survival Convolutional Neural Network (SSCNN) 3 Experiments 3.1 Implementation Details 3.2 Results and Discussion 4 Conclusion References Beyond Non-maximum Suppression - Detecting Lesions in Digital Breast Tomosynthesis Volumes 1 Introduction 2 The DBTex Grand Challenge 3 Data Sets 4 Method 4.1 Models Architecture 4.2 Ensemble 5 Results 5.1 Results on DBTex Dataset 5.2 Ablation Study Results 5.3 Comparison with Prior Art 6 Discussion References A Structural Causal Model for MR Images of Multiple Sclerosis 1 Introduction 2 Related Work 3 Methods 3.1 Background on Structural Causal Models 3.2 High-Resolution Image Generation in Variational Autoencoders 3.3 Learning and Inference in the SCM and VAE 4 Experiments 4.1 Data 4.2 SCM for MS 4.3 Small Images, Large Range 4.4 Large Images, Small Range 5 Discussion References EMA: Auditing Data Removal from Trained Models 1 Introduction 2 Preliminary 2.1 Problem Formulation 2.2 Previous Method 3 The Proposed Method 3.1 Membership Inference Attack 3.2 Ensembled Membership Auditing (EMA) 4 Experiments 4.1 Benchmark Datasets (MNIST and SVHN) 4.2 Chest X-Ray Datasets 5 Conclusion References AnaXNet: Anatomy Aware Multi-label Finding Classification in Chest X-Ray 1 Introduction 2 Methodology 3 Experiments 3.1 Dataset 3.2 Baselines 3.3 Implementation Details 3.4 Results and Evaluation 4 Conclusion References Projection-Wise Disentangling for Fair and Interpretable Representation Learning: Application to 3D Facial Shape Analysis 1 Introduction 2 Methods 2.1 Projecting Features onto a Direction in the Latent Space 2.2 Implementation in the Auto-encoder 2.3 Projection-Wise Disentangling for Interpretation 3 Experiments 3.1 Dataset and Tasks 3.2 Implementation Details 3.3 Evaluation Metrics 4 Results 4.1 Phenotype Prediction for Gender, BMI, and Height 4.2 Prediction on Maternal Alcohol Consumption During Pregnancy 5 Discussion and Conclusion References Attention-Based Multi-scale Gated Recurrent Encoder with Novel Correlation Loss for COVID-19 Progression Prediction 1 Introduction 2 Methodology 2.1 Overview 2.2 Patch Extraction 2.3 Feature Extraction 2.4 Encoder 2.5 Decoder 3 Experimental Design 3.1 Dataset Description 3.2 Implementation Details 4 Results 5 Conclusion References Correction to: Transfer Learning of Deep Spatiotemporal Networks to Model Arbitrarily Long Videos of Seizures Correction to: Chapter “Transfer Learning of Deep Spatiotemporal Networks to Model Arbitrarily Long Videos of Seizures” in: M. de Bruijne et al. (Eds.): Medical Image Computing and Computer Assisted Intervention – MICCAI 2021, LNCS 12905, https://doi.org/10.1007/978-3-030-87240-3_32 Author Index The eight-volume set LNCS 12901, 12902, 12903, 12904, 12905, 12906, 12907, and 12908 constitutes the refereed proceedings of the 24th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2021, held in Strasbourg, France, in September/October 2021.* The 542 revised full papers presented were carefully reviewed and selected from 1809 submissions in a double-blind review process. The papers are organized in the following topical sections: Part I: image segmentation Part II: machine learning - self-supervised learning; machine learning - semi-supervised learning; and machine learning - weakly supervised learning Part III: machine learning - advances in machine learning theory; machine learning - domain adaptation; machine learning - federated learning; machine learning - interpretability / explainability; and machine learning - uncertainty Part IV: image registration; image-guided interventions and surgery; surgical data science; surgical planning and simulation; surgical skill and work flow analysis; and surgical visualization and mixed, augmented and virtual reality Part V: computer aided diagnosis; integration of imaging with non-imaging biomarkers; and outcome/disease prediction Part VI: image reconstruction; clinical applications - cardiac; and clinical applications - vascular Part VII: clinical applications - abdomen; clinical applications - breast; clinical applications - dermatology; clinical applications - fetal imaging; clinical applications - lung; clinical applications - neuroimaging - brain development; clinical applications - neuroimaging - DWI and tractography; clinical applications - neuroimaging - functional brain networks; clinical applications - neuroimaging - others; and clinical applications - oncology Part VIII: clinical applications - ophthalmology; computational (integrative) pathology; modalities - microscopy; modalities - histopathology; and modalities - ultrasound *The conference was held virtually