Medical Image Computing and Computer Assisted Intervention – MICCAI 2022: 25th International Conference, Singapore, September 18–22, 2022, Proceedings, Part III

✍ Scribed by Linwei Wang; Qi Dou; P. Thomas Fletcher; Stefanie Speidel; Shuo Li

• Publisher: Springer Nature
• Year: 2022
• Tongue: English
• Leaves: 832
• Category: Library

✦ Synopsis

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.

✦ Table of Contents

Preface
Organization
Contents – Part III
Breast Imaging
Multi-view Local Co-occurrence and Global Consistency Learning Improve Mammogram Classification Generalisation
1 Introduction
2 Proposed Method
2.1 Multi-view Local Co-occurrence and Global Consistency Learning
3 Experimental Results
3.1 Datasets
3.2 Implementation Details
3.3 Results
3.4 Ablation Study
4 Conclusion
References
Knowledge Distillation to Ensemble Global and Interpretable Prototype-Based Mammogram Classification Models
1 Introduction
2 Proposed Method
2.1 BRAIxProtoPNet++
3 Experimental Results
3.1 Dataset
3.2 Experimental Setup
3.3 Results
4 Conclusion
References
Deep is a Luxury We Don't Have
1 Introduction
2 HCT: High Resolution Convolutional Transformer
2.1 Attention-Convolution (AC) Block
2.2 Efficient Attention
2.3 The HCT Architecture
3 Experiments
4 Ablation Study
5 Conclusion
References
PD-DWI: Predicting Response to Neoadjuvant Chemotherapy in Invasive Breast Cancer with Physiologically-Decomposed Diffusion-Weighted MRI Machine-Learning Model
1 Introduction
2 Method
2.1 Patient Cohort
2.2 Physiological Decomposition of Clinical Breast DWI
2.3 Machine Learning Model
2.4 Evaluation Methodology
3 Results
4 Conclusions
References
Transformer Based Multi-view Network for Mammographic Image Classification
1 Introduction
2 Related Work
3 Methods
3.1 Baseline Models
3.2 Proposed Models
4 Experiments and Results
4.1 Data
4.2 Setting
4.3 Results and Analysis
5 Discussion and Conclusion
References
Intra-class Contrastive Learning Improves Computer Aided Diagnosis of Breast Cancer in Mammography
1 Introduction
2 Method
2.1 Multi-task Learning
2.2 Contrastive Losses
2.3 Hard Negative Sampling
3 Experiments
3.1 Dataset and Metrics
3.2 Implementation Detail
3.3 Result and Analysis
4 Conclusion
References
Colonoscopy
BoxPolyp: Boost Generalized Polyp Segmentation Using Extra Coarse Bounding Box Annotations
1 Introduction
2 Related Work
3 Method
3.1 Fusion Filter Sampling
3.2 Image Consistency (IC) Loss
3.3 Loss Function
4 Experiments
4.1 Datasets and Training Settings
4.2 Quantitative Comparison
4.3 Visual Comparison
4.4 Ablation Study
5 Conclusion
References
FFCNet: Fourier Transform-Based Frequency Learning and Complex Convolutional Network for Colon Disease Classification
1 Introduction
2 Methodology
2.1 Patch Shuffling Module (PSM)
2.2 Frequency-Domain Complex Network
3 Experiments and Results
4 Conclusion
References
Contrastive Transformer-Based Multiple Instance Learning for Weakly Supervised Polyp Frame Detection
1 Introduction and Background
2 Method
2.1 Convolutional Transformer MIL Network
2.2 Transformer-Based MIL Training
3 Experiments and Results
3.1 Dataset
3.2 Implementation Details
3.3 Evaluation on Polyp Frame Detection
3.4 Ablation Study
4 Conclusion
References
Lesion-Aware Dynamic Kernel for Polyp Segmentation
1 Introduction
2 Methodology
2.1 Lesion-Aware Dynamic Kernel
2.2 Attention Modules
3 Experiments
3.1 Datasets
3.2 Implementation Details and Evaluation Metrics
3.3 Experiments on the Public Polyp Benchmarks
3.4 Experiments on the Collected Large-Scale Polyp Dataset
3.5 Ablation Study
4 Conclusion
References
Stepwise Feature Fusion: Local Guides Global
1 Introduction
2 Methodology
2.1 Transformer Encoder
2.2 Aggregate Local and Global Features Stepwise (PLD)
2.3 Stepwise Segmentation Transformer
3 Experiments
3.1 Experimental Setup
3.2 Results
3.3 Ablation Study
4 Conclusions
References
Stay Focused - Enhancing Model Interpretability Through Guided Feature Training
1 Introduction
2 Methods
2.1 Dataset Prepossessing
2.2 Guided Feature Training
2.3 Evaluation
3 Results
3.1 Dataset
3.2 Instrument Segmentation
3.3 Instrument Prediction
3.4 Model Interpretability
3.5 Fake Images
4 Conclusion
References
On the Uncertain Single-View Depths in Colonoscopies
1 Introduction
2 Preliminaries and Related Work
3 Supervised Learning Using Deep Ensembles
4 Self-supervised Learning Using Deep Ensembles
5 Teacher-Student with Uncertain Teacher
6 Experimental Results
7 Conclusions
References
Toward Clinically Assisted Colorectal Polyp Recognition via Structured Cross-Modal Representation Consistency
1 Introduction
2 Related Work
3 Method
3.1 Problem Formulation and Framework Overview
3.2 Shared Transformer Block
3.3 Cross-Modal Global Alignment
3.4 Spatial Attention Module
4 Experiments and Results
4.1 Dataset
4.2 Implementation Details
4.3 Results and Analysis
5 Conclusion
References
TGANet: Text-Guided Attention for Improved Polyp Segmentation
1 Introduction
2 Method
2.1 Encoder Module
2.2 Feature Enhancement Module
2.3 Label Attention
2.4 Decoder
2.5 Multi-scale Feature Aggregation
2.6 Joint Loss Optimization
3 Experiments and Results
3.1 Datasets
3.2 Implementation Details
3.3 Results
3.4 Ablation Study
4 Conclusion
References
Computer Aided Diagnosis
SATr: Slice Attention with Transformer for Universal Lesion Detection
1 Introduction
2 Method
2.1 Multi-slice-Input Backbone
2.2 Slice Attention Transformer
2.3 Hybrid Network
3 Experiments
3.1 Dataset and Setting
3.2 Lesion Detection Performance
3.3 Ablation Study
4 Conclusion
References
MAL: Multi-modal Attention Learning for Tumor Diagnosis Based on Bipartite Graph and Multiple Branches
1 Introduction
2 Proposed Framework
3 Experiment and Results
4 Conclusion
References
Optimal Transport Based Ordinal Pattern Tree Kernel for Brain Disease Diagnosis
1 Introduction
2 Methods
2.1 Data and Preprocessing
2.2 OT-OPT Kernel
2.3 OT-OPT Kernel Based Learning
3 Experiments
3.1 Experimental Setup
3.2 Classification Results
3.3 Discriminative OPTs
3.4 Regression Results
4 Conclusion
References
Dynamic Bank Learning for Semi-supervised Federated Image Diagnosis with Class Imbalance
1 Introduction
2 Method
2.1 Overview of the imFed-Semi Framework
2.2 Dynamic Bank Construction for Unlabeled Clients
2.3 Model Training on Unlabeled Client
3 Experiments
3.1 Dataset and Experiment Setup
3.2 Comparison with State-of-the-Art Methods
3.3 Analytical Studies on Key Components of the Approach
4 Conclusion
References
Coronary R-CNN: Vessel-Wise Method for Coronary Artery Lesion Detection and Analysis in Coronary CT Angiography
1 Introduction
2 Method
2.1 Detection Module for Proposals Localization
2.2 Multi-head Analysis Module for Plaque Analyzing
3 Data
4 Experiments and Results
4.1 Ablation Experiments
4.2 Comparison with Other Methods
5 Conclusion
References
Flat-Aware Cross-Stage Distilled Framework for Imbalanced Medical Image Classification
1 Introduction
2 Method
2.1 Problem Formulation
2.2 Two-Stage Imbalanced Classification Framework
2.3 Flattening Local Minima
2.4 Cross-Stage Distillation
3 Experiments
3.1 Dataset and Evaluation Metrics
3.2 Implementation Details
3.3 Comparison to the State-of-the-Arts
3.4 Ablation Studies
4 Conclusion
References
CephalFormer: Incorporating Global Structure Constraint into Visual Features for General Cephalometric Landmark Detection
1 Introduction
2 Methods
2.1 CephalFormer for Coarse Landmark Detection
2.2 Fine-Scale Coordinate Refinement
3 Experiments
3.1 Settings
3.2 Comparisons with State-of-the-Art Methods
3.3 Ablation Study
4 Conclusion
References
ORF-Net: Deep Omni-Supervised Rib Fracture Detection from Chest CT Scans
1 Introduction
2 Method
2.1 Problem Statement and Formulation
2.2 Omni-Supervision with Dynamic Label Assignment
2.3 Optimization and Implementation Details
3 Experiments and Results
3.1 Dataset and Evaluation Metrics
3.2 Comparison with State-of-the-Art Methods
4 Conclusion
References
Point Beyond Class: A Benchmark for Weakly Semi-supervised Abnormality Localization in Chest X-Rays
1 Introduction
2 Revisit of Point DETR
3 Method: Point Beyond Class
4 Experiments
4.1 Ablation Study
4.2 Performance Benchmark
5 Conclusion
A The Appendix of Point Beyond Class
References
Did You Get What You Paid For? Rethinking Annotation Cost of Deep Learning Based Computer Aided Detection in Chest Radiographs
1 Introduction
2 Methods
2.1 Common Architecture
2.2 Annotation Granularity Definitions
3 Experiments
3.1 Data and Settings
3.2 Classification
3.3 Segmentation
4 Discussion and Conclusions
References
Context-Aware Transformers for Spinal Cancer Detection and Radiological Grading
1 Introduction
2 Related Work
3 Spinal Context Transformer
4 Detecting Spine Cancer Using Radiological Reports
5 Radiological Grading of Spinal Degenerative Changes
6 Experimental Results
7 Conclusion
References
End-to-End Evidential-Efficient Net for Radiomics Analysis of Brain MRI to Predict Oncogene Expression and Overall Survival
1 Introduction
1.1 Importance of MGMT Prediction
1.2 Importance of OS Prediction
1.3 Challenges
1.4 Contribution of This Work
2 Methods
2.1 Overview
2.2 Evidential Regression
2.3 Model
3 Experiments and Results
3.1 Datasets and Implementation
3.2 Results of Predicting MGMT
3.3 Results of Predicting OS
4 Conclusions
References
Denoising of 3D MR Images Using a Voxel-Wise Hybrid Residual MLP-CNN Model to Improve Small Lesion Diagnostic Confidence
1 Introduction
2 Methods
3 Experiments and Discussion
3.1 Dataset
3.2 Training Details
3.3 Evaluation Methods
3.4 Results and Discussion
3.5 Conclusions
References
ULTRA: Uncertainty-Aware Label Distribution Learning for Breast Tumor Cellularity Assessment
1 Introduction
2 Methods
2.1 Label Distribution Generation
2.2 Multi-branch Feature Fusion
2.3 Label Distribution Learning
3 Experiments and Results
3.1 Materials and Evaluation Metrics
3.2 Implementation Details
3.3 Experimental Results and Discussion
4 Conclusion
References
Test-Time Adaptation with Calibration of Medical Image Classification Nets for Label Distribution Shift
1 Introduction
2 Method
2.1 Problem Formulation of Label Distribution Shift
2.2 Learning Diverse Classifiers via Distribution Calibration
2.3 Test-Time Adaptation for Dynamic Classifier Aggregation
3 Experiment
3.1 Dataset and Experimental Setup
3.2 Experimental Results
4 Conclusion
References
Interaction-Oriented Feature Decomposition for Medical Image Lesion Detection
1 Introduction
2 Proposed Method
2.1 IOFD Architecture
2.2 Global Context Embedding (GCE)
2.3 Global Context Cross Attention (GCCA)
3 Experiments
3.1 Datasets, Evaluations and Implementation Details
3.2 Ablation Study
3.3 Comparison Experiments
4 Conclusion
References
Prototype Learning of Inter-network Connectivity for ASD Diagnosis and Personalized Analysis
1 Introduction
2 Dataset and Preprocessing
3 Proposed Method
3.1 Intra-network and Inter-network Encoder
3.2 Prototype-Based Classifier
3.3 Decoder
4 Results
4.1 Experimental Settings
4.2 Experimental Results
5 Personalized FC Analysis
6 Conclusion
References
Effective Opportunistic Esophageal Cancer Screening Using Noncontrast CT Imaging
1 Introduction
2 Methods
3 Experiments
4 Conclusion
References
Joint Prediction of Meningioma Grade and Brain Invasion via Task-Aware Contrastive Learning
1 Introduction
2 Method
2.1 Feature Disentanglement
2.2 Task-Aware Contrastive Learning
2.3 Overall Loss Functions and Training Strategy
3 Experiments
4 Conclusion
References
Residual Wavelon Convolutional Networks for Characterization of Disease Response on MRI
1 Introduction
2 Previous Work and Novel Contributions
3 Methodology
4 Experimental Design
4.1 Data Description
4.2 Model Implementation and Evaluation
5 Experimental Results and Discussion
5.1 Experiment 1: Optimizing Skip Connection Weights in RWCN
5.2 Experiment 2: Comparison of RWCNs Against Alternatives
6 Concluding Remarks
References
Self-Ensembling Vision Transformer (SEViT) for Robust Medical Image Classification
1 Introduction
2 Proposed Method
2.1 Self-Ensembling for Robust Classification
2.2 Adversarial Sample Detection
3 Result and Discussion
3.1 Results on Adversarial Robustness of SEViT
3.2 Results on Adversarial Sample Detection
4 Conclusion
References
A Penalty Approach for Normalizing Feature Distributions to Build Confounder-Free Models
1 Introduction
2 Methodology
2.1 MDN Review
2.2 PMDN: MDN as a Bi-level Optimization
3 Experiments
3.1 Synthetic Dataset Experiments
3.2 Multi-label Multi-site MRI Dataset Experiments
4 Conclusion
References
mulEEG: A Multi-view Representation Learning on EEG Signals
1 Introduction
2 Methods
3 Experiments
4 Results
5 Conclusion
References
Automatic Detection of Steatosis in Ultrasound Images with Comparative Visual Labeling
1 Introduction and Related Works
2 Methods
2.1 Datasets and Labeling
2.2 Converting Comparative Visual Labels to Pathological Scores
2.3 Classification and Regression
3 Experimental Results
3.1 Label Quality
3.2 CNN Model Performance
4 Conclusions
References
Automation of Clinical Measurements on Radiographs of Children's Hips
1 Introduction
2 Methods
2.1 Images
2.2 Manual Measurements
2.3 Automatic Search Models
2.4 Automatic Measurements
3 Results
3.1 Manual Measurements
3.2 Landmark Detection
3.3 Automatic Measurements
4 Discussion and Conclusions
References
Supervised Contrastive Learning to Classify Paranasal Anomalies in the Maxillary Sinus
1 Introduction
2 Method
2.1 Dataset
3 Experiments and Discussion
3.1 Implementation Details
3.2 Evaluation of Learnt Representations
3.3 Label Efficient Representation Learning
4 Conclusion
References
Show, Attend and Detect: Towards Fine-Grained Assessment of Abdominal Aortic Calcification on Vertebral Fracture Assessment Scans
1 Introduction
2 Related Work
2.1 Kauppila Scale and AAC Classes
2.2 Automatic AAC Classification
3 Proposed Framework
4 Experiments
4.1 Dataset
4.2 Pre-processing
4.3 Model and Training Parameters
4.4 Evaluation
4.5 Results and Discussion
5 Conclusion
References
Overlooked Trustworthiness of Saliency Maps
1 Introduction
2 Method
2.1 Relevance of Saliency Maps
2.2 Resistance of Saliency Maps
3 Experiments
3.1 Experimental Details
3.2 Saliency Maps May Not Correlate with Model Outputs
3.3 Saliency Maps Can be Tampered Without Changing Prediction
3.4 The Existence of Transferable Saliency Map Attacks
4 Conclusion
References
Flexible Sampling for Long-Tailed Skin Lesion Classification
1 Introduction
2 Methodology
2.1 SSL Pre-training for Balanced Representations
2.2 Sample Anchor Points Using the Prototype
2.3 Curriculum Sampling Module
3 Experiments
3.1 Datasets
3.2 Implementation Details
3.3 Comparison Study
3.4 Ablation Study
4 Conclusion
References
A Novel Deep Learning System for Breast Lesion Risk Stratification in Ultrasound Images
1 Introduction
2 Methodology
2.1 Multitask Label Generating Architecture
2.2 Consistency Supervision Mechanism
2.3 Cross-Class Loss Function
3 Experiments
4 Conclusion
References
Deep Treatment Response Assessment and Prediction of Colorectal Cancer Liver Metastases
1 Introduction
2 Methods
2.1 Liver Metastasis Segmentation
2.2 Treatment Response Assessment
2.3 Treatment Response Prediction
3 Experiments and Results
3.1 Imaging Datasets
3.2 Implementation Details
3.3 Performance
3.4 Ablation Study
4 Conclusion
References
CACTUSS: Common Anatomical CT-US Space for US Examinations
1 Introduction
2 Method and Experimental Setup
2.1 Data
2.2 Training
2.3 Evaluation Metrics
2.4 Experiments
3 Results and Discussion
3.1 Discussion
4 Conclusion
References
INSightR-Net: Interpretable Neural Network for Regression Using Similarity-Based Comparisons to Prototypical Examples
1 Introduction
2 Methods
3 Experimental Setup
4 Results and Discussion
5 Conclusion
References
Disentangle Then Calibrate: Selective Treasure Sharing for Generalized Rare Disease Diagnosis
1 Introduction
2 Method
2.1 Gradient-Induced Disentanglement (GID)
2.2 Distribution-Targeted Calibration (DTC)
3 Experiments
4 Conclusion
References
Learning Robust Representation for Joint Grading of Ophthalmic Diseases via Adaptive Curriculum and Feature Disentanglement
1 Introduction
2 Methodology
3 Experiments
4 Conclusion
References
Spatiotemporal Attention for Early Prediction of Hepatocellular Carcinoma Based on Longitudinal Ultrasound Images
1 Introduction
2 Methodology
2.1 Problem Formulation
2.2 CNN Image Feature Extractor
2.3 ROI Attention Block
2.4 Transformer for HCC Prediction
3 Experiments
3.1 Experimental Settings
3.2 Evaluations
4 Conclusion
References
NVUM: Non-volatile Unbiased Memory for Robust Medical Image Classification
1 Introduction and Background
2 Method
2.1 Non-volatile Unbiased Memory (NVUM) Training
3 Experiment
3.1 Experiments and Results
3.2 Ablation Study
4 Conclusions and Future Work
References
Local Graph Fusion of Multi-view MR Images for Knee Osteoarthritis Diagnosis
1 Introduction
2 Method
2.1 Knee Graph Construction
2.2 Local Fusion Network (LFN)
2.3 Graph Transformer Network (GTN)
3 Experimental Results
3.1 Data and Experimental Settings
3.2 Comparison with State-of-the-Art Methods
3.3 Contribution of Multiple Views
3.4 Ablation Study on the Design of LGF-Net
4 Conclusion
References
DeepCRC: Colorectum and Colorectal Cancer Segmentation in CT Scans via Deep Colorectal Coordinate Transform
1 Introduction
2 Method
2.1 Colorectal Coordinate Transform
2.2 Network Training
2.3 Network Architecture
3 Experiments
3.1 Dataset and Annotation
3.2 Implementation Details
3.3 Results
4 Conclusion
References
Graph Convolutional Network with Probabilistic Spatial Regression: Application to Craniofacial Landmark Detection from 3D Photogrammetry
1 Introduction
2 Materials and Methods
2.1 Data
2.2 Graph Construction
2.3 Graph Convolutional Network
3 Experiments and Results
3.1 Training Details
3.2 Performance Evaluation
4 Discussion
5 Conclusion
References
Dual-Distribution Discrepancy for Anomaly Detection in Chest X-Rays
1 Introduction
2 Method
2.1 Dual-Distribution Modeling
2.2 Dual-Distribution Discrepancy-Based Anomaly Scores
2.3 Uncertainty-Refined Dual-Distribution Discrepancy
3 Experiments
3.1 Datasets
3.2 Implementation Details
3.3 Ablation Study
3.4 Comparison with State-of-the-Art Methods
4 Conclusion
References
Skin Lesion Recognition with Class-Hierarchy Regularized Hyperbolic Embeddings
1 Introduction
2 Method
2.1 Hyperbolic Geometry
2.2 Hyperbolic Prototype Network for Image Classification
2.3 Incorporating Constraint of Class Relations
3 Experiment and Results
3.1 Dataset and Implementation
3.2 Evaluation Metrics
3.3 Quantitative Results
3.4 Visualization Results
4 Conclusion
References
Reinforcement Learning Driven Intra-modal and Inter-modal Representation Learning for 3D Medical Image Classification
1 Introduction
2 Method
2.1 RL-Based Intra-modality Learning
2.2 RL-Based Inter-modality Learning
2.3 Reward Function and Training Procedure
3 Experiment
3.1 Experiment Setup
3.2 Experimental Results
4 Conclusion
References
A New Dataset and a Baseline Model for Breast Lesion Detection in Ultrasound Videos
1 Introduction
2 Method
2.1 Inter-video Fusion Module
2.2 Intra-video Fusion Module
2.3 Our Network
3 Experiments and Results
3.1 Comparisons with State-of-the-Arts
3.2 Ablation Study
4 Conclusion
References
RemixFormer: A Transformer Model for Precision Skin Tumor Differential Diagnosis via Multi-modal Imaging and Non-imaging Data
1 Introduction
2 Method
2.1 Unified Transformer Backbone
2.2 Disease-Wise Pairing
2.3 Cross-Modality Fusion
3 Experiment and Results
4 Conclusion
References
Building Brains: Subvolume Recombination for Data Augmentation in Large Vessel Occlusion Detection
1 Introduction
2 Methods
2.1 Data
2.2 Preprocessing
2.3 Hemisphere Recombination
2.4 Recombination of ICA and MCA Subvolumes
2.5 Architectures
2.6 Experiments
3 Results
4 Conclusion
References
Hybrid Spatio-Temporal Transformer Network for Predicting Ischemic Stroke Lesion Outcomes from 4D CT Perfusion Imaging
1 Introduction
2 Methods
3 Experiments and Results
4 Conclusion
References
A Medical Semantic-Assisted Transformer for Radiographic Report Generation
1 Introduction
2 Methodology
2.1 Memory-Augmented Sparse Attention of High-Order Interaction
2.2 Image Encoder
2.3 Medical Concepts Generation Network
2.4 Report Decoder
3 Experiments
3.1 Data Collection
3.2 Experimental Settings
3.3 Results and Discussion
4 Conclusions
References
Personalized Diagnostic Tool for Thyroid Cancer Classification Using Multi-view Ultrasound
1 Introduction
2 Methods
2.1 Multi-view Classification Model
2.2 Personalized Weighting Generation
2.3 View-Aware Contrastive Loss
3 Materials and Experiments
4 Results and Discussion
5 Conclusions
References
Morphology-Aware Interactive Keypoint Estimation
1 Introduction
2 Methodology
2.1 Interactive Keypoint Estimation
2.2 Interaction-Guided Gating Network
2.3 Morphology-Aware Loss
3 Experiments
4 Conclusion
References
GazeRadar: A Gaze and Radiomics-Guided Disease Localization Framework
1 Introduction
2 Methodology
2.1 Teacher Block
2.2 Student Block
2.3 Radiomics-Visual Attention Loss
3 Datasets and Environment
4 Results and Discussion
5 Conclusion
References
Deep Reinforcement Learning for Detection of Inner Ear Abnormal Anatomy in Computed Tomography
1 Introduction
2 Data
3 Methods
4 Results
5 Conclusion
References
Vision-Language Contrastive Learning Approach to Robust Automatic Placenta Analysis Using Photographic Images
1 Introduction
2 Method
2.1 Problem Formulation
2.2 Hypothetical Cause of Feature Suppression Problem
2.3 Negative Logarithmic Hyperbolic Cosine Similarity
2.4 Sub-feature Comparison
3 Dataset
4 Experiments
4.1 Training and Testing
4.2 Results and Discussion
5 Conclusions and Future Work
References
Multi-modal Hypergraph Diffusion Network with Dual Prior for Alzheimer Classification
1 Introduction
2 Proposed Framework
2.1 Hypergraph Embeddings and Construction
2.2 Dynamically Adjusted Hypergraph Diffusion Model
3 Experimental Results
4 Conclusion
References
Federated Medical Image Analysis with Virtual Sample Synthesis
1 Introduction
2 Our Method
2.1 Preliminaries and Notations
2.2 Virtual Samples Synthesis for Federated Learning
2.3 Overall Training Algorithm
3 Experiments
3.1 Datasets and Federated Learning Settings
3.2 Compared Methods and Evaluation Metrics
3.3 Experimental Results
3.4 Sensitivity Studies
4 Conclusions
References
Unsupervised Cross-Domain Feature Extraction for Single Blood Cell Image Classification
1 Introduction
2 Methodology
2.1 Domain Adaptation
3 Evaluation
3.1 Datasets
3.2 Implementation Details
3.3 Single Cell Detection by Mask R-CNN
3.4 Evaluation
4 Discussion and Conclusion
References
Attentional Generative Multimodal Network for Neonatal Postoperative Pain Estimation
1 Introduction
2 Related Works
3 Methodology
4 Experimental Setup and Results
5 Conclusion
References
Deep Learning Based Modality-Independent Intracranial Aneurysm Detection
1 Introduction
1.1 Background
1.2 Contributions
2 Data and Methods
2.1 Vessel Extraction
2.2 Aneurysm Detection
3 Experiments and Results
4 Discussion and Conclusion
References
LIDP: A Lung Image Dataset with Pathological Information for Lung Cancer Screening
1 Introduction
2 Related Work
2.1 Dataset
2.2 LIDC-IDRI Based DNN Models for Benign and Malignant Classification
3 LIDP Dataset
3.1 Data Collection and Annotation
3.2 Dataset Properties
3.3 Comparison of LIDP and LIDC-IDRI
4 Experimental Setup
4.1 Data Preprocessing
4.2 Training Method and Parameter Setting
5 Evaluation Experiment
5.1 The Generalization of SOTA Models Trained on LIDC-IDRI
5.2 The Performance of SOTA Model Trained on LIDP Dataset
5.3 The Exploration of LIDP Classification Model Baseline
6 Conclusion
References
Moving from 2D to 3D: Volumetric Medical Image Classification for Rectal Cancer Staging
1 Introduction
2 Materials and Methods
2.1 Dataset and Preprocessing
2.2 2D CNN vs. 3D CNN for Anisotropic Rectal Volume Analysis
2.3 Supplementary Objective Function
2.4 Depth Aggregation Function
2.5 Training and Testing
3 Results
3.1 2D CNN vs. 3D CNN for Anisotropic Rectal Volume Analysis
3.2 Supplementary Objective Function
3.3 Evaluation on Aggregation Functions
3.4 Performance Comparison with Radiologists
4 Conclusion
References
Author Index