Dual-Kernel Adapter: Expanding Spatial Horizons for Data-Constrained Medical Image Analysis

2026-02-21

AI Summary

This work addresses the severe performance degradation of conventional adapters in medical image tasks under extremely limited data, where their effective receptive field (ERF) drastically shrinks—often yielding results inferior to linear probing. The study is the first to uncover this failure mechanism and introduces Dual-Kernel Adapter, a lightweight dual-branch architecture that integrates large and small convolutional kernels to simultaneously capture fine-grained local details and broad spatial context, thereby enhancing the model’s spatial awareness. This module can be efficiently incorporated into pre-trained foundation models for fine-tuning and consistently outperforms existing adapter methods across diverse medical image classification and segmentation benchmarks, regardless of data scarcity or abundance.

Technology Category

• Machine Learning: Kernel Methods
• Computer Vision: Adversarial Attacks & Robustness
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

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
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Adapters have become a widely adopted strategy for efficient fine-tuning of large pretrained models, particularly in resource-constrained settings. However, their performance under extreme data scarcity, common in medical imaging due to high annotation costs, privacy regulations, and fragmented datasets, remains underexplored. In this work, we present the first comprehensive study of adapter-based fine-tuning for large pretrained models in low-data medical imaging scenarios. We find that, contrary to their promise, conventional adapters can degrade performance under severe data constraints, performing even worse than simple linear probing when trained on less than 1% of the corresponding training data. Through systematic analysis, we identify a sharp reduction in Effective Receptive Field (ERF) as a key factor behind this degradation. Motivated by these findings, we propose the Dual-Kernel Adapter (DKA), a lightweight module that expands spatial context via large-kernel convolutions while preserving local detail with small-kernel counterparts. Extensive experiments across diverse classification and segmentation benchmarks show that DKA significantly outperforms existing adapter methods, establishing new leading results in both data-constrained and data-rich regimes.

Problem

• medical image analysis
• data scarcity
• adapter
• low-data regime
• fine-tuning

Innovation

• Dual-Kernel Adapter
• Effective Receptive Field
• data-constrained medical imaging
• parameter-efficient fine-tuning
• large-kernel convolution