MDD-UNet: Domain Adaptation for Medical Image Segmentation with Theoretical Guarantees, a Proof of Concept
This work addresses domain adaptation for medical image segmentation, offering a proof of concept that is incremental but opens new avenues for studying adaptation in models with large hypothesis spaces.
The authors tackled the problem of domain shift in medical image segmentation by proposing MDD-UNet, an unsupervised domain adaptation framework for U-Nets with theoretical guarantees, which improved performance over the standard U-Net on 11 out of 12 dataset combinations for hippocampus segmentation.
The current state-of-the art techniques for image segmentation are often based on U-Net architectures, a U-shaped encoder-decoder networks with skip connections. Despite the powerful performance, the architecture often does not perform well when used on data which has different characteristics than the data it was trained on. Many techniques for improving performance in the presence of domain shift have been developed, however typically only have loose connections to the theory of domain adaption. In this work, we propose an unsupervised domain adaptation framework for U-Nets with theoretical guarantees based on the Margin Disparity Discrepancy [1] called the MDD-UNet. We evaluate the proposed technique on the task of hippocampus segmentation, and find that the MDD-UNet is able to learn features which are domain-invariant with no knowledge about the labels in the target domain. The MDD-UNet improves performance over the standard U-Net on 11 out of 12 combinations of datasets. This work serves as a proof of concept by demonstrating an improvement on the U-Net in it's standard form without modern enhancements, which opens up a new avenue of studying domain adaptation for models with very large hypothesis spaces from both methodological and practical perspectives. Code is available at https://github.com/asbjrnmunk/mdd-unet.