The choice of tracking features in ultrasound-based strain imaging analysis

In order to quantify the left ventricular (LV) strain variation from echocardiographic data, the use of the underlying radio frequency (RF) signal is sometimes proposed to provide higher estimation precision than the B-scan signal. In this paper, we argue that this approach is appropriate only when the deformation between two successive frames is very small. In the analysis of large deformation (such as cardiac deformation), the advantage of using RF signal disappears. Simulation and phantom studies show that the B-scan signal is more robust to frame-to-frame decorrelation than the RF signal when the correlation coefficient of the corresponding RF signal is lower than a threshold value, which is typically equal to 0.985.