STRUDEL: Self-Training with Uncertainty Dependent Label Refinement across Domains

TL;DR

This paper introduces STRUDEL, an unsupervised domain adaptation method for WMH segmentation that uses uncertainty-guided self-training to improve pseudo label quality and model performance across datasets.

Contribution

The paper presents a novel uncertainty-dependent label refinement technique integrated into self-training for improved domain adaptation in medical image segmentation.

Findings

STRUDEL significantly outperforms standard self-training methods.

Incorporating existing robust segmentation outputs enhances pseudo label quality.

The approach improves WMH segmentation accuracy across different datasets.

Abstract

We propose an unsupervised domain adaptation (UDA) approach for white matter hyperintensity (WMH) segmentation, which uses Self-Training with Uncertainty DEpendent Label refinement (STRUDEL). Self-training has recently been introduced as a highly effective method for UDA, which is based on self-generated pseudo labels. However, pseudo labels can be very noisy and therefore deteriorate model performance. We propose to predict the uncertainty of pseudo labels and integrate it in the training process with an uncertainty-guided loss function to highlight labels with high certainty. STRUDEL is further improved by incorporating the segmentation output of an existing method in the pseudo label generation that showed high robustness for WMH segmentation. In our experiments, we evaluate STRUDEL with a standard U-Net and a modified network with a higher receptive field. Our results on WMH segmentation across datasets demonstrate the significant improvement of STRUDEL with respect to standard self-training.