Active few-shot segmentation by reinforcing data selection

2026-07-24

AI Summary

This work addresses the critical yet overlooked issue of support set selection in few-shot medical image segmentation, where the choice of support samples significantly impacts model adaptation performance. Existing approaches typically neglect sample complementarity and resort to independent scoring strategies, limiting their effectiveness. To overcome these limitations, this study proposes the first reinforcement learning–based framework for support set selection, which jointly predicts an optimal support set directly from an unlabeled image pool while explicitly modeling inter-sample complementarity. Experiments on multi-institutional pelvic MRI datasets demonstrate that the proposed method substantially outperforms both random selection and state-of-the-art alternatives, underscoring the pivotal role of support set complementarity in enhancing downstream segmentation performance.

Abstract

Few-shot learning enables medical image segmentation models to adapt to new tasks using only a small number of labelled examples. However, adaptation performance depends strongly on which examples are selected for the support set. Effective support sets should capture relevant variation within the target domain and be informative for adaptation, with constituent samples providing complementary information. Despite this, existing active data selection approaches largely prioritise samples individually and do not explicitly account for interactions between examples. In this work, we propose a reinforcement learning framework for support-set selection in few-shot medical image segmentation, enabling support sets to be optimised jointly rather than through independent sample scoring. Given a pool of unlabelled candidate images, an agent directly predicts a support set that maximises downstream segmentation performance. Experiments on a cross-institutional pelvic MRI dataset demonstrate improvements over random selection and current state-of-the-art methods. Our findings highlight the importance of support-set complementarity for effective adaptation and demonstrate the potential of reinforcement learning for optimising adaptation sets.

Problem

• few-shot segmentation
• active data selection
• support set
• medical image segmentation
• sample complementarity

Innovation

• few-shot segmentation
• active data selection
• reinforcement learning
• support set optimization
• medical image segmentation