... learning methods for mass detection in digital mammography and analyzed in-depth the sources...

类型: 期刊

作者及作者单位: Lidia Garrucho;Kaisar Kushibar;Socayna Jouide;Oliver Diaz;Laura Igual;Karim Lekadir（Artificial Intelligence in Medicine Lab (BCN-AIM), Faculty of Mathematics and Computer Science, University of Barcelona, Gran Via de les Corts Catalanes 585, Barcelona, 08007, Barcelona, Spain）

期刊名: Artificial Intelligence in Medicine

年份: 2022

卷号: Vol.132

期号: Suppl C

页码: P102386

摘要: ... learning methods for mass detection in digital mammography and analyzed in-depth the sources...

外文摘要: 基于深度学习的计算机辅助检测系统在乳腺癌检测中显示出巨大的潜力。 然而，缺乏人工神经网络的领域泛化是其在不断变化的临床环境中应用的一个重要障碍。 在本研究中，我们探索了数字乳腺X射...

Domain generalization in deep learning based mass detection in mammography: A large-scale multi-center study 1
Introduction 1
Related work 2
Domain generalization in medical imaging 2
Mass detection in FFDM using deep learning 3
Robustness of Transformer-based architectures 3
Transfer learning in breast cancer detection 3
Full-Field Digital Mammography (FFDM) datasets 3
OPTIMAM mammography image database 3
INbreast dataset 3
Breast Cancer Digital Repository (BCDR) 4
Domain shift in mammography 4
Methodology 4
Object detection methods 4
Anchor-based detectors 4
Anchor-free detectors 5
Transformer-based detection models 5
Data preparation and training 5
Single Source Domain Generalization (SSDG) 6
Intensity Scale Standardization 6
Data augmentation methods for domain generalization 6
Synthesizing novel domains using MixStyle 6
Image feature extraction backbones 7
Transfer learning on unseen domains 7
Evaluation metrics 7
Experiments and results 7
Performance comparison of mass detection models 7
Single-source domain generalization techniques 7
Computation complexity and performance 8
Detection performance by mass and breast attributes 9
Mass status 9
Mass size 9
Age 9
Breast density 10
Transfer learning on unseen domains 10
Discussion 11
Conclusion 12
Declaration of competing interest 12
Acknowledgments 12
References 12

基于深度学习的计算机辅助检测系统在乳腺癌检测中显示出巨大的潜力。 然而,缺乏人工神经网络的领域泛化是其在不断变化的临床环境中应用的一个重要障碍。 在本研究中,我们探索了数字乳腺X射...

2024-12-16