Strain Estimator in Ultrasound Elastography Using Multiple Frames

Abstract

Accuracy of the motion estimation, higher resolution, robustness to noise and decorrelation arising from out-of-plane motion, faster execution speeds, preciseness to be integrated into subsequent quantitative analysis algorithms are the desired outcomes of a well-established strain estimation algorithm suited for ultrasound elastography. Two strain estimation approaches based on multiple frames that integrate optical flow have been presented to address the primary limitations of existing methodologies. These two proposed techniques attempt to increase the accuracy of displacement estimates, increase spatial resolution over existing methods, and be immune to noise caused by signal decorrelation arising due to the rapid oscillation of RF data and error propagation. Both the techniques use the envelope form of RF data, i.e., B-mode data which is less impacted by signal decorrelation and produce axial frame-sized strain estimates. The proposed techniques were tested on finite element simulation data and in-vivo/patient data. Our techniques statistically outperformed other well-established strain estimators in terms of performance, i.e., SNR and CNR. Even at higher applied compressions like 8% and 16% our proposed methods are highly effective and usable where the other algorithms fail terribly. Spatial resolution was increased by a factor of ~30 than the other algorithms making elastograms similar in dimensions to the frame size.