Improving tissue elastography image quality using multi-position shear wave stimulation and LMS/AHI-based filtering

• In this study, authors proposed a novel approach that stimulates to generate shear waves at multiple-positions alternate on the skin. At each stimulus position, the elasticity and viscosity images are estimated. Next, combines the resulting viscoelastic images into a composite image. This solution improves significantly the obtained-image clarity, particularly in deeper tissue regions, by overcoming attenuation of shear wave. • Least Mean Squares (LMS) adaptive filter is used to suppress random noise on shear wave. The Algebraic Helmholtz Inversion (AHI) algorithm is applied to estimate elasticity and viscosity from the filtered shear wave velocity data. The estimated -viscoelasticity of all pixels gather together to viscoelasticity images. After than a median filter is used to reduce speckle artifacts. • The quality of the estimated images was quantitatively evaluated using the Q-index and Root Mean Square Error (RMSE). Results showed that the composite images from dual-position stimulation achieved higher Q-index values and lower RMSE values compared to images from single-position stimulation. This prove better structural similarity to ideal images and improved signìicantly estimation accuracy. Shear Wave Elastography (SWE) enables non-invasive assessment of tissue stiffness and viscosity, offering high diagnostic value in detecting conditions such as liver cirrhosis and cancers. However, image quality in SWE is often degraded by random noise, speckle artifacts, and signal attenuation, especially in deep tissues. To address these challenges, we propose an image enhancement framework that combines multi-position shear wave stimulation with Least Mean Squares (LMS) filtering and The algebraic Helmholtz inversion (AHI) algorithm. Multiple viscoelasticity images are reconstructed from different stimulation positions and then fused to reinforce consistent features while suppressing noise. A median filter is subsequently applied to further reduce speckle noise. Experimental results, evaluated using Q-index and RMSE, demonstrate a significant improvement in image clarity—particularly in low-SNR regions far from the stimulation source. The proposed method improves viscoelasticity image quality and has the potential to enhance diagnostic confidence in clinical SWE applications.