Medical image computing and computer assisted intervention -- MICCAI 2022 : 25th International Conference, Singapore, September 18-22, 2022, Proceedings. Part IV

The eight-volume set LNCS 13431, 13432, 13433, 13434, 13435, 13436, 13437, and 13438 constitutes the refereed proceedings of the 25th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2022, which was held in Singapore in September 2022. The 574 revised full papers presented were carefully reviewed and selected from 1831 submissions in a double-blind review process. The papers are organized in the following topical sections: Part I: Brain development and atlases; DWI and tractography; functional brain networks; neuroimaging; heart and lung imaging; dermatology; Part II: Computational (integrative) pathology; computational anatomy and physiology; ophthalmology; fetal imaging; Part III: Breast imaging; colonoscopy; computer aided diagnosis; Part IV: Microscopic image analysis; positron emission tomography; ultrasound imaging; video data analysis; image segmentation I; Part V: Image segmentation II; integration of imaging with non-imaging biomarkers; Part VI: Image registration; image reconstruction; Part VII: Image-Guided interventions and surgery; outcome and disease prediction; surgical data science; surgical planning and simulation; machine learning – domain adaptation and generalization; Part VIII: Machine learning – weakly-supervised learning; machine learning – model interpretation; machine learning – uncertainty; machine learning theory and methodologies. Preface Organization Contents – Part IV Microscopic Image Analysis An End-to-End Combinatorial Optimization Method for R-band Chromosome Recognition with Grouping Guided Attention 1 Introduction 2 Method 2.1 Overview 2.2 Grouping Guided Feature Interaction Module 2.3 Deep Assignment Module 3 Experiments 3.1 Datasets and Implementation Details 3.2 Results on Normal Karyotypes 3.3 Results on Karyotypes with Numerical Abnormalities 4 Conclusion References Efficient Biomedical Instance Segmentation via Knowledge Distillation 1 Introduction 2 Methodology 3 Experiments 3.1 Datasets and Metrics 3.2 Implementation Details 3.3 Experimental Results 4 Conclusion References Tracking by Weakly-Supervised Learning and Graph Optimization for Whole-Embryo C. elegans lineages 1 Introduction 2 Method 3 Results 4 Conclusion References Mask Rearranging Data Augmentation for 3D Mitochondria Segmentation 1 Introduction 2 Method 2.1 3D EM Image Generator 2.2 3D Mask Layout Generator 3 Experiments 3.1 Dataset and Experiment Settings 3.2 Experiments and Results 4 Conclusions References Semi-supervised Learning for Nerve Segmentation in Corneal Confocal Microscope Photography 1 Introduction 2 The Proposed Method 2.1 Pre-training 2.2 Model Fine-Tuning 2.3 Self-training 3 Evaluations 3.1 Data Set 3.2 Experimental Setup 3.3 Ablation Study 3.4 Comparisons with Semi-supervised Methods 4 Conclusions References Implicit Neural Representations for Generative Modeling of Living Cell Shapes 1 Introduction 2 Method 2.1 Learning a Latent Space of Shapes 2.2 Neural Network Architecture 2.3 Data 3 Experiments and Results 3.1 Reconstruction of Cell Sequences 3.2 Generating New Cell Sequences 3.3 Temporal Interpolation 3.4 Generating Benchmarking Datasets 4 Discussion and Conclusion References Trichomonas Vaginalis Segmentation in Microscope Images 1 Introduction 2 Proposed Dataset 2.1 Data Collection 2.2 Dataset Features 3 Method 3.1 Overview 3.2 High-Resolution Fusion Module 3.3 Foreground-Background Attention Module 4 Experiments and Results 4.1 Experimental Settings 4.2 Comparison with State-of-the-Art 4.3 Ablation Study 5 Conclusion References NerveFormer: A Cross-Sample Aggregation Network for Corneal Nerve Segmentation 1 Introduction 2 Proposed Method 2.1 CNN Encoder 2.2 Deformable and External Attention Module (DEAM) 2.3 CNN Decoder 3 Experiments 3.1 Datasets and Implementation Details 3.2 Comparison with the State-of-the-Art Methods 3.3 Ablation Study 3.4 Conclusion References Domain Adaptive Mitochondria Segmentation via Enforcing Inter-Section Consistency 1 Introduction 2 Related Works 3 Method 4 Experiments 5 Conclusion References DeStripe: A Self2Self Spatio-Spectral Graph Neural Network with Unfolded Hessian for Stripe Artifact Removal in Light-Sheet Microscopy 1 Introduction 2 Methods 2.1 Detecting of Corruption in Fourier Space 2.2 Formulating Stripe Removal as a Deep Unfolding Framework 2.3 Graph-Based Fourier Recovery Network G( ) 2.4 Unfolded Hessian Prior for Structure Preservation H( ) 2.5 Self2Self Denoising Loss Formulation 2.6 Competitive Methods 3 Results and Discussion 3.1 Evaluation on LSFM Images with Synthetic Stripe Artifact 3.2 Evaluation on LSFM Images with Real Stripe Artifact 4 Conclusion References End-to-End Cell Recognition by Point Annotation 1 Introduction 2 Methods 3 Experiments 3.1 Dataset Description and Experimental Settings 3.2 Experimental Results 4 Conclusion References ChrSNet: Chromosome Straightening Using Self-attention Guided Networks 1 Introduction 2 Methods 2.1 Curved Chromosome Synthesizer 2.2 ChrSNet: Chromosome Straightening Networks 3 Experiments 4 Conclusion References Region Proposal Rectification Towards Robust Instance Segmentation of Biological Images 1 Introduction 2 Method 2.1 Baseline Framework 2.2 Region Proposal Rectification 3 Experiment 3.1 Datasets 3.2 Experimental Details 3.3 Evaluation Metric 4 Results and Discussion 5 Conclusion References DeepMIF: Deep Learning Based Cell Profiling for Multispectral Immunofluorescence Images with Graphical User Interface 1 Introduction 2 Materials 3 Methodology 3.1 Cell Detection and Classification on Deconvoluted Images 3.2 Markers Co-expression Identification 3.3 DeepMIF Graphical User Interface 4 Results and Discussion References Capturing Shape Information with Multi-scale Topological Loss Terms for 3D Reconstruction 1 Introduction 2 Related Work 3 Our Method: A Topology-Aware Loss 3.1 Assessing the Topology of Volumes 3.2 Loss Term Construction 4 Experiments 4.1 Training and Evaluation 4.2 Results 5 Discussion References Positron Emission Tomography MCP-Net: Inter-frame Motion Correction with Patlak Regularization for Whole-body Dynamic PET 1 Introduction 2 Methods 2.1 Dataset and Pre-processing 2.2 Proposed Network 2.3 Training Details and Baseline Comparison 2.4 Evaluation Metrics 3 Results 3.1 Motion Simulation Test 3.2 Qualitative Analysis 3.3 Quantitative Analysis 4 Conclusion References PET Denoising and Uncertainty Estimation Based on NVAE Model Using Quantile Regression Loss 1 Introduction 2 Related Work 2.1 Variational Autoencoder (VAE) 2.2 Nouveau Variational Autoencoder (NVAE) 3 Methods 3.1 Overview 3.2 PET Image Denoising 3.3 Quantile Regression Loss 4 Experiment 4.1 Dataset 4.2 Data Analysis 5 Results 6 Discussion 7 Conclusion References TransEM: Residual Swin-Transformer Based Regularized PET Image Reconstruction 1 Introduction 2 Methods and Materials 2.1 Problem Formulation 2.2 Residual Swin-Transformer Regularizer 2.3 Implementation Details and Reference Methods 3 Experiment and Results 3.1 Experimental Evaluation 3.2 Results 3.3 Robustness Analysis 3.4 Ablation Study and Discussion 4 Conclusions References Supervised Deep Learning for Head Motion Correction in PET 1 Introduction 2 Methods 2.1 Data 2.2 Motion Correction Network Structure 2.3 Network Training Strategy 2.4 Motion Correction Inference 3 Results 3.1 Single Subject Pilot Experiments 3.2 Multi-subject Experiments 4 Discussion and Conclusion References Ultrasound Imaging Adaptive 3D Localization of 2D Freehand Ultrasound Brain Images 1 Introduction 2 Methods 2.1 Problem Setup 2.2 Training with Sampled 2D Slices from 3D Volumes 2.3 Fine-tuning with 2D Ultrasound Images 2.4 Inference 3 Experimental Design 4 Results and Discussion 4.1 Volume-Sampled Images 4.2 Native Freehand Images 5 Conclusion References Physically Inspired Constraint for Unsupervised Regularized Ultrasound Elastography 1 Introduction 2 Method 2.1 Hook's Law and Poisson's Ratio 2.2 Physically Inspired ConsTraint for Unsupervised Regularized Elastography (PICTURE) 2.3 Unsupervised Training 2.4 Data Collection 2.5 Network Architecture and Training Schedule 3 Results 3.1 Experimental Phantom Results 3.2 In Vivo results 3.3 Quantitative Results 4 Conclusions References Towards Unsupervised Ultrasound Video Clinical Quality Assessment with Multi-modality Data 1 Introduction 2 Related Work 3 Method 3.1 Model Structure 3.2 Objective Function 4 Experiment and Results 5 Conclusion References Key-frame Guided Network for Thyroid Nodule Recognition Using Ultrasound Videos 1 Introduction 2 Method 3 Experiments and Results 4 Conclusions References Less is More: Adaptive Curriculum Learning for Thyroid Nodule Diagnosis 1 Introduction 2 Related Work 3 Methodology 3.1 Hard Sample Discovery with Confidence Queue 3.2 Model Certainty Estimation with Certainty Queue 3.3 Loss Function of Adaptive Curriculum Learning 4 TNCD: Benchmark for Thyroid Nodule Classification 5 Experiment 5.1 Implementation and Evaluation Metric 5.2 Ablation Study and Schedule Analysis 5.3 Comparison with the State-of-the-arts 6 Conclusion References Localizing the Recurrent Laryngeal Nerve via Ultrasound with a Bayesian Shape Framework 1 Introduction 2 Methods 2.1 Bayesian Shape Alignment 2.2 Locate-Net 3 Experiments 3.1 Dataset and Evaluation Metrics 3.2 Implementation Details 3.3 Comparison Baselines 3.4 Ablation Study 4 Conclusion References Uncertainty-aware Cascade Network for Ultrasound Image Segmentation with Ambiguous Boundary 1 Introduction 2 Methods 2.1 Architecture 2.2 Objective Function 3 Experiments 3.1 Comparison with State-of-the-Arts 3.2 Ablation Study 4 Conclusion References BiometryNet: Landmark-based Fetal Biometry Estimation from Standard Ultrasound Planes 1 Introduction 2 Method 2.1 Landmark Regression Network 2.2 Dynamic Orientation Determination (DOD) 2.3 Scale Recovery 3 Experimental Setup 4 Results 5 Conclusions References Deep Motion Network for Freehand 3D Ultrasound Reconstruction 1 Introduction 2 Methodology 2.1 Temporal and Multi-branch Structure for IMU Fusion 2.2 Multi-modal Online Self-supervised Strategy 3 Experiments 4 Conclusion References Agent with Tangent-Based Formulation and Anatomical Perception for Standard Plane Localization in 3D Ultrasound 1 Introduction 2 Method 2.1 Reinforcement Learning for Plane Localization 2.2 Auxiliary Task of State-Content Similarity Prediction 2.3 Imitation Learning Based Initialization 3 Experimental Result 3.1 Materials and Implementation Details 3.2 Quantitative and Qualitative Analysis 4 Conclusion References Weakly-Supervised High-Fidelity Ultrasound Video Synthesis with Feature Decoupling 1 Introduction 2 Methodology 2.1 Weakly-Supervised Training for Motion Estimation 2.2 Dual-Decoder Generator for Content and Texture Decoupling 2.3 Adversarial Learning and GAN Loss for Sharpness Improvement 3 Experiments and Results 4 Conclusions References Class Impression for Data-Free Incremental Learning 1 Introduction 2 Method 2.1 Class Impression 2.2 Novel Losses 3 Experiments 3.1 Datasets and Experimental Settings 3.2 Comparison with Baselines 3.3 Ablation Studies 4 Conclusion References Simultaneous Bone and Shadow Segmentation Network Using Task Correspondence Consistency 1 Introduction 2 Method 2.1 Preliminaries 2.2 Network Architecture 2.3 Cross Task Feature Transfer Block 2.4 Task Correspondence Consistency Loss 3 Experiments and Results 4 Discussion 5 Conclusion References Contrastive Learning for Echocardiographic View Integration 1 Introduction 2 Methods 2.1 Volume Contrastive Network 2.2 Volume Contrastive Losses 3 Experiments and Results 3.1 Experiment Settings 3.2 Results 4 Conclusion References BabyNet: Residual Transformer Module for Birth Weight Prediction on Fetal Ultrasound Video 1 Introduction 2 Method 2.1 Feature Extraction 2.2 Residual Transformer Module 2.3 3D Multi-Head Self-Attention 3 Experiments 4 Discussion 5 Conclusions References EchoGNN: Explainable Ejection Fraction Estimation with Graph Neural Networks 1 Introduction 2 Related Work 3 Methodology 3.1 EchoGNN Architecture 4 Experiments 4.1 Dataset 4.2 Implementation 4.3 Results and Discussion 4.4 Ablation Study 5 Limitations 6 Conclusion References EchoCoTr: Estimation of the Left Ventricular Ejection Fraction from Spatiotemporal Echocardiography 1 Introduction 2 Related Works 3 Methods 3.1 Frames Sampling 3.2 Architecture Overview 3.3 Existing Methods for LVEF Estimation 4 Experiments 4.1 Datasets 4.2 Experimental Setup 5 Results 6 Discussion 7 Conclusion References Light-weight Spatio-Temporal Graphs for Segmentation and Ejection Fraction Prediction in Cardiac Ultrasound 1 Introduction 1.1 Prior Work 1.2 Contributions: 2 Method 2.1 Building Blocks 2.2 EchoGraphs - Left Ventricle Segmentation and EF Prediction 3 Experiments and Results 3.1 Dataset 3.2 Segmentation 3.3 EF Prediction 4 Discussion and Conclusion References Rethinking Breast Lesion Segmentation in Ultrasound: A New Video Dataset and A Baseline Network 1 Introduction 2 Method 2.1 Query and Memory Encoder 2.2 Parallel Spatial Temporal Transformer 2.3 Decoder 2.4 Dynamic Memory Selection 3 Experiments 3.1 Dataset and Implementation 3.2 Comparison with State-of-the-Art Methods 3.3 Ablation Study 4 Conclusion References MIRST-DM: Multi-instance RST with Drop-Max Layer for Robust Classification of Breast Cancer 1 Introduction 2 Related Works 2.1 Adversarial Attacks and Defenses 2.2 Breast Ultrasound Image Classification 3 Proposed Method 3.1 Multi-instance RST 3.2 Drop-Max Layer 3.3 SimCLR Pretraining 4 Experimental Results 4.1 Experiment Setup 4.2 The Effectiveness of Multiple-Instance RST 4.3 The Effectiveness of the SimCLR Pretrained Model 4.4 The Effectiveness of the Drop-Max Layer 5 Conclusion References Towards Confident Detection of Prostate Cancer Using High Resolution Micro-ultrasound 1 Introduction 2 Materials and Methods 2.1 Data 2.2 Methodology 3 Experiments and Results 3.1 Effect of Co-teaching 3.2 Comparison of Uncertainty Methods 3.3 Model Demonstration 4 Conclusion References Video Data Analysis Unsupervised Contrastive Learning of Image Representations from Ultrasound Videos with Hard Negative Mining 1 Introduction 2 Datasets 2.1 In-house US Dataset for Gallbladder Cancer 2.2 Public Lung US Dataset for COVID-19 3 Our Method 4 Experiments and Results 5 Conclusion References An Advanced Deep Learning Framework for Video-Based Diagnosis of ASD 1 Introduction 2 Methodology 2.1 Video Acquisition of Children 2.2 Advanced Deep Learning Framework 3 Experiments and Results 3.1 Implementation Details 3.2 Experimental Results 3.3 Analysis and Discussion 4 Conclusion References Automating Blastocyst Formation and Quality Prediction in Time-Lapse Imaging with Adaptive Key Frame Selection 1 Introduction 2 Method 2.1 Policy Network 2.2 Prediction Network 2.3 Loss Function 3 Experimental Results 4 Conclusions References Semi-supervised Spatial Temporal Attention Network for Video Polyp Segmentation 1 Introduction 2 Method 2.1 Temporal Local Context Attention 2.2 Proximity Frame Temporal-Spatial Attention 2.3 Loss Function 2.4 Training Flow 3 Experiments 3.1 Datasets and Implementation 3.2 Qualitative Evaluation 3.3 Quantitative Evaluation 3.4 Ablation Study 4 Conclusion References Geometric Constraints for Self-supervised Monocular Depth Estimation on Laparoscopic Images with Dual-task Consistency 1 Introduction 2 Method 2.1 Self-supervised Learning 2.2 Dual-task Consistency Loss and Weight Mask 3 Experiments and Results 3.1 Datasets and Evaluation Metrics 3.2 Implementation Details 3.3 Comparison Results 3.4 Ablation Study 4 Discussion and Conclusions References Recurrent Implicit Neural Graph for Deformable Tracking in Endoscopic Videos 1 Introduction 2 Related Work 3 Methods 4 Experiments 5 Results 6 Conclusion References Pose-Based Tremor Classification for Parkinson's Disease Diagnosis from Video 1 Introduction 2 Method 2.1 Pose Extraction 2.2 Classification Network 3 Experiments 4 Conclusion References Image Segmentation I Neural Annotation Refinement: Development of a New 3D Dataset for Adrenal Gland Analysis 1 Introduction 2 Method 2.1 Deep Implicit Surfaces 2.2 Neural Annotation Refinement 3 Datasets 3.1 Distorting a Golden Standard Segmentation Dataset 3.2 ALAN Dataset: A New 3D Dataset for Adrenal Gland Analysis 4 Experiments 4.1 Quantitative Experiments on Distorted Golden Standards 4.2 Adrenal Diagnosis on the Repaired ALAN Dataset 5 Conclusion References Few-shot Medical Image Segmentation Regularized with Self-reference and Contrastive Learning*-10pt 1 Introduction 2 Methodology 2.1 Problem Setting 2.2 Local Prototype-Based Segmentation 2.3 Self-reference Regularization 2.4 Contrastive Learning 2.5 Superpixel-Based Self-supervised Learning 3 Experiments 3.1 Experimental Setup 3.2 Comparison with the State-of-the-Art (SOTA) Methods 3.3 Ablation Studies 4 Conclusion References Shape-Aware Weakly/Semi-Supervised Optic Disc and Cup Segmentation with Regional/Marginal Consistency 1 Introduction 2 Methods 2.1 Modified Signed Distance Function (mSDF) 2.2 Dual Consistency Regularisation of Semi-Supervision 2.3 Differentiable vCDR estimation of Weakly Supervision 3 Experiments 3.1 Datasets and Implementation Details 4 Results 5 Conclusion References Accurate and Robust Lesion RECIST Diameter Prediction and Segmentation with Transformers 1 Introduction 2 Method 2.1 The Architecture of MeaFormer 2.2 Model Optimization 3 Experiments 4 Conclusions References DeSD: Self-Supervised Learning with Deep Self-Distillation for 3D Medical Image Segmentation 1 Introduction 2 Method 2.1 Overview 2.2 Deep Self-Distillation 2.3 Downstream Transfer Learning 2.4 Architecture Details 3 Experiments and Results 3.1 Datasets and Evaluation Metrics 3.2 Implementation Details 3.3 Results 4 Conclusion References Self-supervised 3D Anatomy Segmentation Using Self-distilled Masked Image Transformer (SMIT) 1 Introduction 2 Method 3 Experiments and Results 4 Discussion and Conclusion References DeepRecon: Joint 2D Cardiac Segmentation and 3D Volume Reconstruction via a Structure-Specific Generative Method 1 Introduction 2 Method 2.1 DeepRecon Architecture 2.2 Learning of DeepRecon 2.3 3D Reconstruction and 4D Motion Adaptation 3 Experiments 3.1 Latent-space-based 2D Segmentation 3.2 3D Volume Reconstruction 3.3 Motion Pattern Adaptation 4 Conclusion References Online Easy Example Mining for Weakly-Supervised Gland Segmentation from Histology Images 1 Introduction 2 Method 2.1 Overall Framework 2.2 Online Easy Example Mining 2.3 Network Training 3 Experiments 3.1 Dataset 3.2 Compare with State-of-the-Arts 3.3 Ablation Study 4 Conclusion References Joint Class-Affinity Loss Correction for Robust Medical Image Segmentation with Noisy Labels 1 Introduction 2 Joint Class-Affinity Segmentation Framework 2.1 Differentiated Affinity Reasoning (DAR) 2.2 Class-Affinity Loss Correction (CALC) 3 Experiments 4 Conclusion References Task-Relevant Feature Replenishment for Cross-Centre Polyp Segmentation 1 Introduction 2 Methodology 2.1 Domain-Invariant Feature Decomposition (DIFD) 2.2 Task-Relevant Feature Replenishment 2.3 Polyp-Aware Adversarial Learning (PAAL) 2.4 Overall Network Training 3 Experiments 3.1 Experimental Settings 3.2 Comparison with State-of-the-Arts 3.3 Ablation Study 4 Conclusion References Vol2Flow: Segment 3D Volumes Using a Sequence of Registration Flows 1 Introduction 1.1 Related Works 2 Method 2.1 Vol2Flow Network and Self-Supervised Learning 2.2 Mask Propagation 3 Experimental Setup 4 Results and Discussion 5 Conclusion References Parameter-Free Latent Space Transformer for Zero-Shot Bidirectional Cross-modality Liver Segmentation 1 Introduction 2 Method 3 Experiments and Results 4 Conclusion References Using Guided Self-Attention with Local Information for Polyp Segmentation 1 Introduction 2 Method 3 Experiments 3.1 Experiments on Polyp Segmentation 3.2 Ablation Study 4 Conclusion References Momentum Contrastive Voxel-Wise Representation Learning for Semi-supervised Volumetric Medical Image Segmentation 1 Introduction 1.1 Overview 1.2 Unsupervised Contrastive Learning 2 Experiments 3 Conclusion A Hardness-aware Property of the Contrastive Losses References Context-Aware Voxel-Wise Contrastive Learning for Label Efficient Multi-organ Segmentation 1 Introduction 2 Method 2.1 Supervised Losses for Labeled Voxels 2.2 Context-Aware Contrastive Learning Loss for Unlabeled Voxels 2.3 Overall Loss Function 2.4 Implementation Details 3 Experiments 3.1 Dataset 3.2 Comparing to State-of-the-Art Methods 3.3 Ablation Study 4 Conclusion References Vector Quantisation for Robust Segmentation*-10pt 1 Introduction 1.1 Contribution 2 Methods 2.1 Robustness and Network Assumptions 2.2 Quantisation for Robustness 2.3 Perturbation Bounds 2.4 Implementation Details and Data 3 Experiments 3.1 Codebook Study 3.2 Domain Shift Study 3.3 Perturbation Study 4 Conclusion References A Hybrid Propagation Network for Interactive Volumetric Image Segmentation 1 Introduction 2 Methodology 2.1 Overview 2.2 Volume Propagation Network 2.3 Slice Propagation Network 2.4 Implementation 3 Experiments 3.1 Datasets and Experimental Setup 3.2 Comparison with Previous Methods 3.3 Ablation Study 3.4 User Study 4 Conclusion References SelfMix: A Self-adaptive Data Augmentation Method for Lesion Segmentation*-8pt 1 Introduction 2 Method 2.1 Self-adaptive Data Augmentation 3 Experiments 3.1 Datasets and Implementation Details 3.2 Experimental Results 3.3 Effectiveness Analysis Using SelfMix: 3.4 Relationship with Mixup, CutMix and CarveMix 4 Conclusion References Bi-directional Encoding for Explicit Centerline Segmentation by Fully-Convolutional Networks 1 Introduction 2 Method 3 Experiments 3.1 Data 3.2 Encoding Configuration 3.3 Results 3.4 Ablation Study 4 Discussion 5 Conclusions References Transforming the Interactive Segmentation for Medical Imaging 1 Introduction 2 Methods 2.1 Problem Scenario 2.2 Encoder (ENC) 2.3 Refinement (REF) 3 Experiments 3.1 Datasets 3.2 Evaluation Metrics 3.3 Implementation Details 4 Results 4.1 User Interactions Simulation 4.2 Comparisons with State-of-the-Art 4.3 Ablation Study 4.4 Visualization of Results 4.5 Visualization of the Interaction Process 5 Conclusion References Learning Incrementally to Segment Multiple Organs in a CT Image 1 Introduction 2 Related Work 3 Method 3.1 IL for MOS 3.2 Memory Module 4 Experiments 4.1 Setup 4.2 Results and Discussions 5 Conclusion References Harnessing Deep Bladder Tumor Segmentation with Logical Clinical Knowledge 1 Introduction 2 Method 2.1 Constructing Logic Rules for Bladder Tumor Localization 2.2 Embedding Logic Rules into Latent Features 2.3 Optimization of Segmentation Network with Logic Rules 3 Experiments 3.1 Test of Segmentation Enhancement by Logic Rules 3.2 Comparison with Other Segmentation Methods 4 Conclusion References Test-Time Adaptation with Shape Moments for Image Segmentation 1 Introduction 2 Method 3 Experiments 3.1 Test-time Adaptation with Shape Descriptors 3.2 Results and Discussion 4 Conclusion References Author Index