Towards Practical Algorithm Selection for Unsupervised Domain Adaptation in Medical Imaging

2026-07-30

AI Summary

This work addresses the challenge of model selection in unsupervised domain adaptation for medical imaging, where the absence of labels in the target domain hinders effective algorithm and hyperparameter tuning. The authors propose a novel, label-free joint model selection criterion that aggregates candidate models across diverse algorithms and hyperparameter configurations by leveraging multiple unlabeled signals. This approach constructs a two-stage consistency-based reference prediction and selects the model whose predictions align most closely with this reference for deployment. To the best of our knowledge, this is the first method to enable unified selection of both algorithms and hyperparameters without requiring target-domain labels. Evaluated across eight medical imaging datasets and seven clinical transfer scenarios, the proposed method consistently outperforms existing approaches, demonstrating strong generalization and stability.

Abstract

Numerous unsupervised domain adaptation (UDA) algori-thms exist, but for clinical practice, selecting the best-suited one along with proper hyperparameters often remains unclear, as the unlabeled deployment (target) domain prevents direct evaluation. We propose a label-free criterion that jointly selects the algorithm and hyperparameters for UDA. Given a pool of candidate models from multiple algorithms trained with different hyperparameters, our approach scores each candidate against an agreement reference, and selects the one with the highest score. The agreement reference is constructed in two levels without using target labels. First, we leverage multiple label-free selection signals, using each to nominate a model within every algorithm. Second, the nominated models are aggregated across algorithms to form a reference prediction for each unlabeled target sample. The candidate whose predictions agree most with this reference is then selected for deployment. Experimental results on four brain MRI and four chest X-ray datasets across seven clinically relevant transfer scenarios show that our method achieves better selection performance than other methods and remains effective across different algorithm pools. Our approach takes a step towards practical, label-free algorithm selection for clinical deployment of UDA.

Problem

• unsupervised domain adaptation
• algorithm selection
• medical imaging
• hyperparameter tuning
• label-free evaluation

Innovation

• unsupervised domain adaptation
• algorithm selection
• label-free criterion
• agreement reference
• medical imaging