Wavelet shrinkage-based strain estimation for elastography

Presented in this paper is a wavelet shrinkage-based method for estimating soft tissue strain from displacement images. Through orthogonal wavelet decomposition, the displacement image is first split into a coarse approximate and several detail displacement sub-images in octave sub-bands. Then, all detail displacement sub-images are denoised through wavelet shrinkage. A final strain image is obtained by wavelet reconstruction from all strain sub-images, each of which is the direct gradient of a displacement sub-image. The new strain estimator was evaluated using a simple 2D simulation and an experimental phantom, both containing a stiffer inclusion in a softer background. Simulation and experimental results showed that the new method can be used to enhance the contrast-to-noise ratio (CNR) and signal-to-noise ratio (SNR) of the estimated strain image without sacrificing spatial resolution. Perhaps even more importantly, the new estimator is more time-efficient and holds potential for use in real-time elastography algorithms.