A global strain estimation algorithm for non-invasive vascular ultrasound elastography

Most strain estimation algorithms are window-based. Within a calculation window, cross-correlation or affine estimations are performed. However, there is always a trade-off between the window size, overlap and computation efficiency. We propose a parameterized affine model to estimate pixel-wise strain globally within the framework of the Horn-Schunck optical flow (OF) estimation, which enables to derive a global strain field efficiently without multiple windowed calculations. In addition, properties of the global pixel-wise estimation provide higher strain imaging resolution. Specifically, global strain fields were parameterized with discrete cosine transform (DCT) descriptions. A cost function including an OF constancy term, a smoothness constrain and a nearly incompressibility term was minimized to derive affine strain components (axial and lateral strains and shears), from which principal strains were determined. For the simulation study, the proposed method provided less estimation errors than the window-based Lagrangian speckle model estimator (LSME). The computation time with the proposed method was also reduced by more than 4 times compared with the LSME. For in vitro experiments, the proposed method was found to be able to detect a 1 mm hard inclusion.