Interpolation-Based Regularization for Speed of Sound Estimation in Layered Media

In biological media, acoustic waves propagate at a speed determined by tissue properties. The local estimation of sound velocity is therefore promising for several clinical applications, such as the assessment of liver steatosis. Recently, a method that estimates the local speed of sound in real time has been proposed in the literature. This method uses refraction effects to obtain axial speed of sound profiles. To improve the resolution and robustness of such profiles while preserving the real time capability, we propose a regularization approach that assumes a piecewise constant speed of sound and formulates a simple convex regularization problem based on this hypothesis. We show that this optimization problem can be solved in less than 100ms and that it significantly improves both the resolution and robustness in simulations, in vitro and in vivo conditions.