SA-LAIS-PCGS: Speckle-Aware Restoration and Perceptual Enhancement of Breast Synthetic Aperture Ultrasound Images

This paper proposes a novel two-stage image quality restoration and enhancement framework for speckle-aware noise suppression and perceptual improvement of breast synthetic aperture ultrasound (SA-US) images. The framework integrates a Speckle-Aware Locally Adaptive Intensity Smoother (SA-LAIS) with a Percentile Contrast Gradient Sharpness (PCGS) module. SA-LAIS is designed to attenuate multiplicative speckle noise by adaptively weighting local intensity ratios combined with gradient magnitude information, effectively reducing noise while preserving delicate anatomical structures. The PCGS module subsequently performs percentile-based contrast stretching, gradient-guided sharpness enhancement, and gamma correction to improve perceptual quality. Experiments on breast ultrasound datasets contaminated with simulated speckle and AWGN demonstrate that SA-LAIS-PCGS achieves a favorable trade-off between noise suppression, structure preservation, and visual enhancement. The proposed model is lightweight, interpretable, and readily deployable as a regularizer within computational SA-US imaging pipelines, contributing to improved diagnostic visualization and opening pathways for integration with inverse problem solvers and deep learning-based reconstruction methods.