CLoPA: Continual Low Parameter Adaptation of Interactive Segmentation for Medical Image Annotation

2026-03-06

AI Summary

Existing zero-shot interactive segmentation models struggle to consistently achieve expert-level performance across diverse medical imaging tasks. This work proposes a parameter-efficient continual adaptation strategy that fine-tunes a small subset of model parameters online using an annotation cache, without introducing new parameters or altering the inference pipeline. It is the first to integrate a lightweight continual learning mechanism into interactive medical image annotation, enabling automatic online optimization without human intervention and revealing the relationship between the efficacy of tuning specific parameter groups and task characteristics. Built upon the nnInteractive framework and incorporating instance normalization, hierarchical feature tuning, and a lightweight training-triggering scheduler, the method rapidly attains expert-level segmentation performance across eight tasks from the Medical Segmentation Decathlon, with particularly notable improvements in segmenting complex anatomical structures.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Life-Long and Continual Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Interactive segmentation enables clinicians to guide annotation, but existing zero-shot models like nnInteractive fail to consistently reach expert-level performance across diverse medical imaging tasks. Because annotation campaigns produce a growing stream of task-specific labelled data, online adaptation of the segmentation model is a natural complement to zero-shot inference. We propose CLoPA, a continual adaptation strategy that tunes a small fraction of nnInteractive's parameters on the annotation cache, triggered by lightweight episode scheduling. CLoPA requires no new parameters or changes to the inference pipeline, and operates entirely within the existing annotation workflow. Across eight Medical Segmentation Decathlon tasks spanning diverse anatomical targets and imaging characteristics, CLoPA rapidly elevates performance to expert-level, even for tasks where nnInteractive previously failed, with the majority of gains realised after a single training episode. We show that the benefits of tuning different parameter groups depends on task characteristics and data regimes. Also, that for targets with complex geometries (e.g., hepatic vessels), instance normalisation and low-level feature tuning saturates, suggesting a need for deeper feature-representation alignment in the most challenging scenarios.

Problem

• interactive segmentation
• medical image annotation
• zero-shot models
• continual adaptation
• expert-level performance

Innovation

• continual adaptation
• low-parameter tuning
• interactive segmentation
• medical image annotation
• zero-shot model refinement