EviATTA: Evidential Active Test-Time Adaptation for Medical Segment Anything Models

2026-03-15

AI Summary

This work addresses the unreliability of supervision in medical Segment Anything Models (SAM) during testing due to distribution shifts, as well as the challenges of inaccurate uncertainty estimation and inefficient use of sparse annotations in existing test-time adaptation methods. To this end, it proposes the first active test-time adaptation framework tailored for medical SAM. The approach leverages Dirichlet-based evidential learning to decompose predictive uncertainty, introduces a hierarchical evidence sampling strategy to efficiently select the most informative samples, and incorporates a dual consistency regularization mechanism to enhance model robustness. Extensive experiments across six medical image segmentation datasets demonstrate that with only a minimal number of expert annotations, the method significantly improves adaptation performance and reliability in both batch-level and instance-level scenarios.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Computer Vision: Segmentation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Deploying foundational medical Segment Anything Models (SAMs) via test-time adaptation (TTA) is challenging under large distribution shifts, where test-time supervision is often unreliable. While active test-time adaptation (ATTA) introduces limited expert feedback to improve reliability, existing ATTA methods still suffer from unreliable uncertainty estimation and inefficient utilization of sparse annotations. To address these issues, we propose Evidential Active Test-Time Adaptation (EviATTA), which is, to our knowledge, the first ATTA framework tailored for medical SAMs. Specifically, we adopt the Dirichlet-based Evidential Modeling to decompose overall predictive uncertainty into distribution uncertainty and data uncertainty. Building on this decomposition, we design a Hierarchical Evidential Sampling strategy, where image-wise distribution uncertainty is used to select informative shifted samples, while distance-aware data uncertainty guides sparse pixel annotations to resolve data ambiguities. We further introduce Dual Consistency Regularization, which enforces progressive prompt consistency on sparsely labeled samples to better exploit sparse supervision and applies variational feature consistency on unlabeled samples to stabilize adaptation. Extensive experiments on six medical image segmentation datasets demonstrate that EviATTA consistently improves adaptation reliability with minimal expert feedback under both batch-wise and instance-wise test-time adaptation settings.

Problem

• test-time adaptation
• medical image segmentation
• active learning
• uncertainty estimation
• distribution shift

Innovation

• Evidential Modeling
• Active Test-Time Adaptation
• Uncertainty Decomposition
• Hierarchical Sampling
• Dual Consistency Regularization