Data Augmentation, Labelling, and Imperfections

Name: Data Augmentation, Labelling, and Imperfections | Third MICCAI Workshop, DALI 2023, Held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, October 12, 2023, Proceedings
Brand: Springer
Price: 106.99 EUR
Availability: OnlineOnly

Third MICCAI Workshop, DALI 2023, Held in Conjunction with MICCAI 2023, Vancouver, BC, Canada, October 12, 2023, Proceedings

Yuan Xue Chen Chen Chao Chen Lianrui Zuo Yihao Liu(Editor)

Springer (Publisher)

Published on 26. April 2024

X, 168 pages

E-Book

PDF with digital watermarking

System requirements

978-3-031-58171-7 (ISBN)

€106.99incl. 7% vat

System requirements

for PDF with digital watermarking

E-Book Single Licence

Available for download

Description

More details

Other editions

Content

Intro
Preface
Organization
Contents
URL: Combating Label Noise for Lung Nodule Malignancy Grading
1 Introduction
2 Method
2.1 Problem Definition and Overview
2.2 SCL Stage
2.3 MU Stage
3 Experiments and Results
3.1 Dataset and Experimental Setup
3.2 Comparative Experiments
3.3 Ablation Analysis
4 Conclusion
References
Zero-Shot Learning of Individualized Task Contrast Prediction from Resting-State Functional Connectomes
1 Introduction
2 Methods
3 Experimental Setup
3.1 Data
3.2 OPIC's Training
3.3 Baselines
3.4 Metrics
4 Results
4.1 In-Domain Prediction Quality
4.2 Out-of-Domain Prediction Quality
4.3 New Task Contrast from a Seen Task Group
5 Conclusion
References
Microscopy Image Segmentation via Point and Shape Regularized Data Synthesis
1 Introduction
2 Methods
3 Experiments
4 Discussion
References
A Unified Approach to Learning with Label Noise and Unsupervised Confidence Approximation
1 Introduction
2 Related Work
3 Method
3.1 Noisy Labels and Confidence Score Approximation
3.2 Unsupervised Confidence Approximation Loss
3.3 Unsupervised Confidence Approximation Architecture
3.4 Confidence-Selective Prediction
3.5 Pixelwise UCA
4 Experimental Results
5 Conclusion
References
Transesophageal Echocardiography Generation Using Anatomical Models
1 Introduction
2 Methods
2.1 Pseudo-Image Generation
2.2 Image Synthesis
3 Results and Discussion
4 Conclusion
References
Data Augmentation Based on DiscrimDiff for Histopathology Image Classification
1 Introduction
2 Method
2.1 Synthesizing Histopathology Images Based on Diffusion Model
2.2 Post-discrimination Mechanism for Diffusion
3 Experiments
3.1 Datasets and Implementation
3.2 Result and Discussion
3.3 Visualization of Class-Specific Image Features
4 Conclusion
References
Clinically Focussed Evaluation of Anomaly Detection and Localisation Methods Using Inpatient CT Head Data
1 Introduction
2 Related Work
3 Dataset
4 Anomaly Detection Models
5 Clinical Evaluation Methodology
6 Results
7 Conclusion
References
LesionMix: A Lesion-Level Data Augmentation Method for Medical Image Segmentation
1 Introduction
1.1 Related Works
1.2 Contributions
2 Method
2.1 LesionMix
2.2 Lesion Populating
2.3 Lesion Inpainting
2.4 Lesion Load Distribution
2.5 Properties of LesionMix
3 Experiments
3.1 Data
3.2 Implementation Details
3.3 Results
4 Conclusion
References
Knowledge Graph Embeddings for Multi-lingual Structured Representations of Radiology Reports
1 Introduction
2 Methodology
3 Experimental Setup
4 Results and Discussion
5 Conclusion
References
Modular, Label-Efficient Dataset Generation for Instrument Detection for Robotic Scrub Nurses
1 Introduction
2 Dataset
2.1 Data Acquisition
2.2 Real Multi-instrument Data for Validation and Testing
2.3 Real Single-Instrument Images for Advanced MBOI
3 Experiments
3.1 Model and Hyperparameters
3.2 Synthetic Training Data from MBOI
3.3 Advancing Copy-Paste in MBOI
3.4 Effciency: Performance vs. Invested Resources
4 Results
4.1 Naive Insertion vs. Gaussian Blur and Poisson Blending
4.2 Impact of the Number of SI Images and Training Set Size
4.3 Evaluation of with Other Detectors Under Optimal Conditions
5 Conclusion
References
Adaptive Semi-supervised Segmentation of Brain Vessels with Ambiguous Labels
1 Introduction
2 Methodology
2.1 Preprocessing
2.2 Problem Formulation
2.3 Supervised Learning
2.4 Semi-supervised Learning
3 Experiments and Results
3.1 Datasets
3.2 Experimental Setup
3.3 Evaluation Metrics
3.4 Qualitative Results and Analysis
3.5 Quantitative Results and Analysis
4 Conclusion
References
Proportion Estimation by Masked Learning from Label Proportion
1 Introduction
2 PD-L1 Tumor Proportion Estimation
3 Experiments
4 Conclusion
References
Active Learning Strategies on a Real-World Thyroid Ultrasound Dataset
1 Background
1.1 Active Learning
1.2 Active Learning Applied to Thyroid Ultrasound
2 Materials and Methods
2.1 Image Datasets
2.2 Rigged Draw Strategy
2.3 Supervised and Unsupervised Active Learning Strategies
3 Results
3.1 Supervised Strategies
3.2 Semi-supervised Strategies
4 Discussion
References
A Realistic Collimated X-Ray Image Simulation Pipeline
1 Introduction
2 Methods
2.1 Randomized Collimator Simulation Pipeline
2.2 Experiments
3 Results
3.1 Framework Validation
3.2 Network Evaluation
4 Discussion
References
Masked Conditional Diffusion Models for Image Analysis with Application to Radiographic Diagnosis of Infant Abuse
1 Introduction
2 Methods
2.1 Diffusion Model
2.2 Image Generation via Conditional Diffusion Model
3 Experiments
4 Results and Discussion
5 Conclusions
References
Self-supervised Single-Image Deconvolution with Siamese Neural Networks
1 Introduction and Related Work
2 Methods
3 Experiments
3.1 2D Dataset
3.2 3D Dataset
4 Discussion
5 Conclusion
References
Author Index

System requirements

Save as PDF Copy link into clipboard

Schweitzer Fachinformationen

Data Augmentation, Labelling, and Imperfections

Description

More details

Other editions

Additional editions

Content

System requirements