Prior-guided Hierarchical Instance-pixel Contrastive Learning for Ultrasound Speckle Noise Suppression

TL;DR

This paper introduces a hierarchical contrastive learning framework guided by prior knowledge, combining pixel and instance-level strategies with a Transformer-CNN architecture to effectively suppress speckle noise in ultrasound images while preserving structural details.

Contribution

It proposes a novel prior-guided hierarchical contrastive learning model with a hybrid Transformer-CNN architecture for improved ultrasound denoising.

Findings

Outperforms existing ultrasound denoising methods on public datasets

Enhances local structural consistency through pixel-level contrastive learning

Achieves better preservation of anatomical details with the hybrid architecture

Abstract

Ultrasound denoising is essential for mitigating speckle-induced degradations, thereby enhancing image quality and improving diagnostic reliability. Nevertheless, because speckle patterns inherently encode both texture and fine anatomical details, effectively suppressing noise while preserving structural fidelity remains a significant challenge. In this study, we propose a prior-guided hierarchical instance-pixel contrastive learning model for ultrasound denoising, designed to promote noise-invariant and structure-aware feature representations by maximizing the separability between noisy and clean samples at both pixel and instance levels. Specifically, a statistics-guided pixel-level contrastive learning strategy is introduced to enhance distributional discrepancies between noisy and clean pixels, thereby improving local structural consistency. Concurrently, a memory bank is employed to facilitate instance-level contrastive learning in the feature space, encouraging representations that more faithfully approximate the underlying data distribution. Furthermore, a hybrid Transformer-CNN architecture is adopted, coupling a Transformer-based encoder for global context modeling with a CNN-based decoder optimized for fine-grained anatomical structure restoration, thus enabling complementary exploitation of long-range dependencies and local texture details. Extensive evaluations on two publicly available ultrasound datasets demonstrate that the proposed model consistently outperforms existing methods, confirming its effectiveness and superiority.