Optimization of transmit parameters for two-dimensional cardiac strain estimation with coherent compounding in silico, in vitro, and in vivo

Coherent compounding has been investigated as a possible means of increasing strain estimation accuracy while retaining high frame rate; however, the optimal parameters that yield the best strain estimate have yet to be determined. Three transmit parameters were investigated: the number of transmits, the subaperture size, and the angular aperture. Field II simulations of an annular region mimicking cross-sectional views of the systolic left ventricle were imaged. The highest number of transmits evaluated (15), the smallest subaperture (11 elements), and the narrowest angular aperture (60 degrees) were found to yield the smallest axial and lateral interframe strain errors. A hemi-ellipsoid PVA phantom was inflated and deflated in a water tank to generate wall strain, and the in silico results were corroborated by the phantom results. Finally, a canine infarct model was employed wherein the LAD was ligated, and open chest images of the myocardium were acquired four days later. In silico and in vivo results were used to design an optimized compounding sequence featuring 11 transmits, a 21 element subaperture, and a 90 degrees angular aperture. Based on TTC staining of the excised heart, the optimized sequence was able to distinguish between healthy and infarcted tissue, whereas the single diverging wave sequence incorrectly estimated strain in several regions. This optimization study indicates that a higher number of transmits, a small subaperture, and a narrow angular aperture yield the most accurate strain estimate.