Intro
Preface
Organization
Contents - Part III
Machine Learning - Explainability, Bias, and Uncertainty II
Pre-trained Diffusion Models for Plug-and-Play Medical Image Enhancement
1 Introduction
2 Method
2.1 Denoising Diffusion Probabilistic Models (DDPM) for Unconditional Image Generation
2.2 Image Enhancement with Denoising Algorithm
2.3 Pre-Trained Diffusion Models for Plug-and-play Medical Image Enhancement
3 Experiments
4 Results and Discussion
5 Conclusion
References
GRACE: A Generalized and Personalized Federated Learning Method for Medical Imaging
1 Introduction
2 Method
2.1 Overview of the GPFL Framework
2.2 Local Training Phase: Feature Alignment &
Personalization
2.3 Aggregation Phase: Consistency-Enhanced Re-weighting
3 Experiments
3.1 Dataset and Experimental Setting
3.2 Comparison with SOTA Methods
3.3 Further Analysis
4 Conclusion
References
Chest X-ray Image Classification: A Causal Perspective
1 Introduction
2 Methodology
2.1 A Causal View on CXR Images
2.2 Causal Intervention via Backdoor Adjustment
2.3 Training Object
3 Experiments
3.1 Experimental Setup
3.2 Results and Analysis
4 Conclusion
References
DRMC: A Generalist Model with Dynamic Routing for Multi-center PET Image Synthesis
1 Introduction
2 Method
2.1 Center Interference Issue
2.2 Network Architecture
2.3 Dynamic Routing Strategy
2.4 Loss Function
3 Experiments and Results
3.1 Dataset and Evaluation
3.2 Implementation
3.3 Comparative Experiments
3.4 Ablation Study
4 Conclusion
References
Federated Condition Generalization on Low-dose CT Reconstruction via Cross-domain Learning
1 Introduction
2 Method
2.1 iRadonMAP
2.2 Proposed FedCG Method
3 Experiments
3.1 Dataset.

3.2 Implementation Details
4 Result
4.1 Reuslt on Condition #1
4.2 Result on Condition #2
4.3 Ablation Experiments
5 Conclusion
References
Enabling Geometry Aware Learning Through Differentiable Epipolar View Translation
1 Introduction
2 Methods
3 Experiments
3.1 Model Training
4 Results
5 Discussion and Conclusion
References
Enhance Early Diagnosis Accuracy of Alzheimer's Disease by Elucidating Interactions Between Amyloid Cascade and Tau Propagation
1 Introduction
2 Method
2.1 Reaction-Diffusion Model for Neuro-Dynamics
2.2 Construction on the Interaction Between Tau and Amyloid
2.3 Neural Network Landscape of RDM-Based Dynamic Model
3 Experiments
3.1 Data Description and Experimental Setting
3.2 Ablation Study in Prediction Disease Progression
3.3 Prognosis Accuracies on Forecasting AD Risk
4 Conclusion
References
TauFlowNet: Uncovering Propagation Mechanism of Tau Aggregates by Neural Transport Equation
1 Introduction
2 Methods
2.1 Problem Formulation for Discovering Spreading Flow of Tau Propagation
2.2 TauFlowNet: An Explainable Deep Model Principled with TV-Based Lagrangian Mechanics
3 Experiments
3.1 Evaluate the Prediction Accuracy of Future Tau Accumulation
3.2 Examine Spatiotemporal Patterns of the Spreading Flow of Tau Aggregates
4 Conclusion
References
Uncovering Structural-Functional Coupling Alterations for Neurodegenerative Diseases
1 Introduction
2 Method
2.1 Generalized Kuramoto Model for Coupled Neural Oscillations
2.2 Deep Kuramoto Model for SC-FC Coupling Mechanism
2.3 Novel SC-FC Coupling Biomarkers
3 Experiments
3.1 Validating the Neuroscience Insight of Deep Kuramoto Model
3.2 Evaluation on Empirical Biomarker of SC-FC-META
3.3 Evaluation on SC-FC-Net in Diagnosing AD
4 Conclusion.

4 Conclusions
References
How Reliable are the Metrics Used for Assessing Reliability in Medical Imaging?
1 Introduction
2 Related Work
3 Proposed Methodology
3.1 Proposed Metric: Robust Expected Calibration Error (RECE)
3.2 Proposed Robust Calibration Regularization (RCR) Loss
4 Experiments and Results
5 Conclusion
References
Co-assistant Networks for Label Correction
1 Introduction
2 Methodology
2.1 Noise Detector
2.2 Noise Cleaner
2.3 Objective Function
3 Experiments
3.1 Experimental Settings
3.2 Results and Analysis
3.3 Ablation Study
4 Conclusion
References
M3D-NCA: Robust 3D Segmentation with Built-In Quality Control
1 Introduction
2 Methodology
2.1 M3D-NCA Training Pipeline
2.2 M3D-NCA Core Architecture
2.3 Inherent Quality Control
3 Experimental Results
3.1 Comparison and Ablation
3.2 Automatic Quality Control
4 Conclusion
References
The Role of Subgroup Separability in Group-Fair Medical Image Classification
1 Introduction
2 Related Work
3 The Role of Subgroup Separability
4 Experiments and Results
5 Discussion
References
Mitigating Calibration Bias Without Fixed Attribute Grouping for Improved Fairness in Medical Imaging Analysis
1 Introduction
2 Methodology
2.1 Training Procedure: Two-Stage Method
2.2 Test Time Evaluation on Subgroups of Interest
3 Experiments and Results
3.1 Results, Ablations, and Analysis
4 Conclusions
References
SMRD: SURE-Based Robust MRI Reconstruction with Diffusion Models
1 Introduction
2 Related Work
3 Method
3.1 Accelerated MRI Reconstruction Using Diffusion Models
3.2 Stein's Unbiased Risk Estimator (SURE)
3.3 SURE-Based MRI Reconstruction with Diffusion Models
4 Experiments
5 Results and Discussion
6 Conclusion
References.

Asymmetric Contour Uncertainty Estimation for Medical Image Segmentation
1 Introduction
2 Method
2.1 Contouring Uncertainty
2.2 Visualization of Uncertainty
3 Experimental Setup
3.1 Data
3.2 Implementation Details
3.3 Evaluation Metrics
4 Results
5 Discussion and Conclusion
References
Fourier Test-Time Adaptation with Multi-level Consistency for Robust Classification
1 Introduction
2 Methodology
3 Experimental Results
4 Conclusion
References
A Model-Agnostic Framework for Universal Anomaly Detection of Multi-organ and Multi-modal Images
1 Introduction
2 Methodology
2.1 Framework Overview
2.2 Organ and Modality Classification Constraints
2.3 Center Constraint
2.4 Optimization and Inference
3 Experiments
3.1 Experimental Setting
3.2 Comparison Study
3.3 Ablation Study
4 Conclusion
References
DiMix: Disentangle-and-Mix Based Domain Generalizable Medical Image Segmentation
1 Introduction
2 Methods
2.1 Framework
2.2 Loss Function
3 Experiments and Results
3.1 Setup
3.2 Implementation Details
3.3 Results
4 Conclusion
References
Regular SE(3) Group Convolutions for Volumetric Medical Image Analysis
1 Introduction
2 Literature Overview
3 Separable SE(n) Equivariant Group Convolutions
3.1 Regular Group Convolutions
3.2 Separable SE(n) Group Convolution
4 Experiments and Evaluation
4.1 Evaluation Methodology
4.2 SE(3) Equivariance Performance
4.3 Performance on MedMNIST
4.4 Model Generalization
4.5 Future Work
5 Conclusion
References
Deep Learning-Based Anonymization of Chest Radiographs: A Utility-Preserving Measure for Patient Privacy
1 Introduction
2 Methods
2.1 Data
2.2 PriCheXy-Net: Adversarial Image Anonymization
2.3 Objective Functions
3 Experiments and Results.