Mask2Restore: Self-Supervised Ultrasound Despeckling via Inpainting

2026-09-26

AI Summary

This study addresses the failure of conventional blind-spot networks in ultrasound despeckling due to the spatial correlation of speckle noise and the absence of clean reference images. To overcome these limitations, this work reformulates despeckling as a context inpainting task, introducing a patch-level masking strategy to circumvent multi-pixel correlations while leveraging anatomical priors for detail restoration. Furthermore, a cross-resolution consistency regularization is incorporated to suppress residual biases, enabling fully self-supervised learning. The proposed method significantly improves the trade-off between signal-to-noise ratio enhancement and structural detail preservation, faithfully recovering fine anatomical structures and effectively boosting performance on downstream cardiac segmentation tasks.

Abstract

Medical ultrasound (US) is inherently degraded by speckle, a granular interference pattern that is often treated as a complex form of noise in image restoration. However, unlike random noise, US speckle originates from coherent scattering within tissue and is therefore highly spatially dependent and deterministic under fixed acquisition conditions, making US speckle suppression fundamentally different from natural image denoising. Because speckle-free US targets are unavailable in practice, self-supervised denoising is necessary. Blind-spot networks (BSN) are the dominant self-supervised paradigm for natural images, but their pixel-wise masking strategy assumes spatially independent noise, an assumption poorly matched to US speckle, which is spatially correlated over multiple pixels rather than pixel-wise independent. To address this mismatch, we propose Mask2Restore, a self-supervised US despeckling framework that reformulates despeckling as contextual inpainting with block-wise masking on single noisy images. Unlike pixel-wise BSN masking, block-wise masking addresses this multi-pixel speckle correlation by removing locally correlated speckle neighborhoods and shifting the reconstruction cues used by the network from adjacent speckle correlations to broader anatomical context. We further introduce cross-resolution context regularization (CRCR), which suppresses residual speckle bias by enforcing consistency across multi-resolution predictions. Experiments on simulated and in vivo carotid US, unseen fine-structure cases, and downstream cardiac segmentation demonstrate improved speckle-detail trade-offs, better preservation of fine anatomical structures, and practical value for subsequent image analysis.

Problem

• ultrasound despeckling
• self-supervised learning
• spatially correlated noise
• blind-spot network
• image restoration

Innovation

• Self-supervised despeckling
• Block-wise masking
• Contextual inpainting
• Cross-resolution context regularization
• Blind-spot networks