Bayesian adaptively-weighted ensembles for few-shot abdominal segmentation

2026-08-06

AI Summary

This work addresses the dual challenges of label scarcity and cross-institutional domain shift, which severely limit the performance of existing few-shot segmentation methods and hinder fixed-weight ensembles from adapting to target domains. To overcome these limitations, the authors propose a Bayesian adaptive weighted ensemble framework: multiple few-shot segmentation models are first fine-tuned, and then Bayesian optimization is employed to automatically learn optimal fusion weights on a target-domain validation set, which are subsequently used for query image prediction. This approach represents the first application of Bayesian optimization to ensemble few-shot medical image segmentation, enabling adaptive weight assignment that accounts for both anatomical structures and institutional domain characteristics. Evaluated on the Cross-institution Male Pelvic Structures dataset, the method significantly outperforms single models, fixed ensembles, models trained from scratch, and current state-of-the-art approaches, demonstrating strong potential for clinical deployment.

Abstract

Few-shot learning has emerged as a promising approach for anatomical segmentation when labelled data are scarce. However, different few-shot learning algorithms exhibit complementary strengths and weaknesses, with performance varying across anatomical targets and institutions. Existing few-shot segmentation ensembles, that combine predictions from multiple algorithms, typically employ fixed weighting schemes and therefore cannot adjust model contributions according to the target domain. In this work, we propose a Bayesian adaptively-weighted ensemble framework for segmentation under label scarcity and domain shift. Multiple few-shot segmentation algorithms are first adapted using a small labelled support set. Bayesian optimisation is then used to automatically identify ensemble weights that maximise segmentation performance on a target-domain validation set. The learned weights are subsequently fixed and applied to combine predictions on previously unseen query images from the target domain. The proposed framework is evaluated on the Cross-institution Male Pelvic Structures dataset using held-out anatomical structures and institutions to simulate simultaneous label scarcity and institutional domain shift. Results demonstrate statistically significant improvements over individual few-shot learners, fixed-weight ensembles, training-from-scratch baselines and recent state-of-the-art ensembling approaches. By adapting model contributions to the target anatomy and institutional domain, the proposed framework provides a practical mechanism for deploying segmentation systems to new clinical sites under severe annotation constraints.

Problem

• few-shot segmentation
• domain shift
• anatomical segmentation
• label scarcity
• ensemble learning

Innovation

• Bayesian optimisation
• adaptively-weighted ensemble
• few-shot segmentation
• domain shift
• medical image segmentation