How Far from Clinical Deployment? Evaluating the Complete Unsupervised Domain Adaptation Pipeline in Medical Imaging

2026-08-12

AI Summary

This study addresses a critical challenge in the clinical deployment of unsupervised domain adaptation (UDA): the absence of labeled data in the target domain impedes reliable model selection, thereby limiting real-world performance. For the first time, UDA algorithms and unsupervised model selectors are jointly evaluated as an integrated pipeline across 11 cross-domain scenarios spanning 9 medical imaging datasets, involving 10 UDA methods and 13 selection strategies—yielding over 80,000 models analyzed. The findings reveal that model selection constitutes a key bottleneck in UDA adoption: although high-performing adapted models often exist, current unsupervised selectors struggle to identify them consistently. While ensemble learning and minimal target-domain annotations substantially narrow the gap to oracle-level performance, they do not fully eliminate it.

Abstract

Deploying unsupervised domain adaptation (UDA) in clinical practice requires choosing which algorithm to use and which of its trained models to ship. However, the deployment (target) domain is unlabeled, so models cannot be evaluated directly on it, leaving it unclear which to select. We address this by evaluating the complete UDA pipeline, considering both adaptation and label-free selection together. Our study covers eleven clinically relevant cross-domain scenarios from nine medical imaging datasets, with ten UDA algorithms and 13 label-free selection methods (validators), evaluating over 80,000 trained models in total. By this, we find that a capable adapted model usually exists, but identifying it without target labels is difficult: the validator-selected models leave a large and structural target performance gap to the best available one, with no evaluated validator consistently reliable. Towards closing it, we explore two strategies, ensembling and a small target-labeling budget; both narrow this gap but do not close it entirely. Overall, deployable UDA depends on the complete pipeline; addressing the less explored selection step could bring much of current UDA closer to clinical use.

Problem

• unsupervised domain adaptation
• medical imaging
• model selection
• label-free evaluation
• clinical deployment

Innovation

• unsupervised domain adaptation
• label-free model selection
• medical imaging
• clinical deployment
• domain shift