Elastography

Introduction
Many diseases cause changes in tissue mechanical properties. Current imaging devices such as computed tomography (CT), ultrasound (US) and magnetic resonance imaging (MRI) are not directly capable of measuring the mechanical properties of soft tissue. Elastography as a strain imaging technique has been well established in the literature as a new imaging method. The strain distributions in tissues in response to an external deformation are closely related to the distribution of tissue elasticity. These strain images can give a clear illustration of the underlying tissue stiffness distributions which has been shown to provide useful clinical information.

Static Elastography
Static elastography is performed by (i) obtaining a set of ultrasonic radio frequency echo signals from a target (i.e. pre-compression RF frame), (ii) subjecting the target to a small axial deformation and (iii) obtaining a second set of echo signals (i.e. post-compression RF frame). Motions along the direction of the applied load are estimated by performing piecewise motion estimation on corresponding pairs of signal segments. Once displacements are calculated, strain estimation algorithms can be applied to generate the strain images generally referred to as elastograms.

Real-Time Elastography
In real-time elastography RF frames are acquired continuously and elastograms are generated in real-time by estimating strain between sequential frames similarly to static methods. Fast algorithms are necessary in this kind of imaging. Also strains in-between-frames are small thus sensitivity of the algorithms is also very important.

Implementation on Sonix Systems
1D normalized cross-correlation with adjustable windows size and overlap is used to estimate the motions. In order to speed up the computation, time-delay estimator with prior estimates (TDPE) was employed. TDPE uses dynamic programming to reduce the size of the search region for the cross-correlation. Following the motion estimation, the strains were estimated using an optimized least square strain estimator. The estimated strains were scaled, filtered, color-coded and displayed on the screen beside or superimposed on sonograms.