On Direct Distribution Matching for Adapting Segmentation Networks

2019/04/04 by Georg Pichler, Jose Dolz, Pichler, Georg +6 · 1 citation
Computer Science · Mathematics · Medicine · #Adversarial Robustness in Machine Learning #Adversarial system #Artificial intelligence #COVID-19 diagnosis using AI #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Computer vision #Context (archaeology) #Discriminator #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Geography #Image segmentation #Kernel (algebra) #Machine learning #Matching (statistics) #Mathematics #Minification #Pattern recognition (psychology) #Scale-space segmentation #Segmentation #Stability (learning theory) #cs.CV

paper · pdf · doi:10.48550/arxiv.1904.02657

published in arXiv (Cornell University), 624-637 (Cornell University) · includes appendix; published at MIDL2020: https://2020.midl.io/papers/pichler20.html

openalex publication_date 2019/04/04 · arxiv created 2021/11/24 · arxiv updated 2021/11/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/06

Abstract

Minimization of distribution matching losses is a principled approach to domain adaptation in the context of image classification. However, it is largely overlooked in adapting segmentation networks, which is currently dominated by adversarial models. We propose a class of loss functions, which encourage direct kernel density matching in the network-output space, up to some geometric transformations computed from unlabeled inputs. Rather than using an intermediate domain discriminator, our direct approach unifies distribution matching and segmentation in a single loss. Therefore, it simplifies segmentation adaptation by avoiding extra adversarial steps, while improving both the quality, stability and efficiency of training. We juxtapose our approach to state-of-the-art segmentation adaptation via adversarial training in the network-output space. In the challenging task of adapting brain segmentation across different magnetic resonance images (MRI) modalities, our approach achieves significantly better results both in terms of accuracy and stability.

Citations

• Learning Transferable Features with Deep Adaptation Networks
• Unsupervised Domain Adaptation with Residual Transfer Networks
• FCNs in the Wild: Pixel-level Adversarial and Constraint-based Adaptation
• PnP-AdaNet: Plug-and-Play Adversarial Domain Adaptation Network with a Benchmark at Cross-modality Cardiac Segmentation
• ADVENT: Adversarial Entropy Minimization for Domain Adaptation in Semantic Segmentation

Related

• Unsupervised domain adaptation in brain lesion segmentation with adversarial networks
• MDD-UNet: Domain Adaptation for Medical Image Segmentation with Theoretical Guarantees, a Proof of Concept
• C-MADA: Unsupervised Cross-Modality Adversarial Domain Adaptation framework for medical Image Segmentation
• Adversarial training and dilated convolutions for brain MRI segmentation
• Domain Adaptive Medical Image Segmentation via Adversarial Learning of Disease-Specific Spatial Patterns