Non-linear beamforming approaches for sizing and detecting large calcifications

Abstract

Standard B-mode imaging has poor sensitivity and specificity for detecting kidney stones and consistently overestimates stone size. Because of this, the acoustic shadow produced by the stone and twinkling artifacts seen with color Doppler have been used as substitutes for conventional imaging for stone sizing and detection. However, often neither a shadow nor a color Doppler artifact are present. In this study, the use of several non-linear beamforming strategies was investigated in conjunction with plane wave synthetic focusing (PWSF). These include aperture domain model image reconstruction (ADMIRE), short-lag spatial coherence (SLSC), and a new mid-lag spatial coherence (MLSC) method designed specifically for kidney stone detection but not sizing. Evaluations of all four methods were performed in vitro and ex vivo. For the in vitro evaluation, various sized kidney stones (n=8 with width 9.88±5.96mm) were placed on top of a gelatin phantom doped with graphite, which served as a platform and provided a diffuse scattering background for comparisons. The stones were imaged at a depth of 4 cm and 8 cm. An ex vivo evaluation was also performed where several stones were implanted into pig kidneys. The pig kidneys were immersed in water for imaging. The in vitro sizing errors for all stones at both depths for PWSF, ADMIRE, SLSC, and MLSC were 0.89±0.74mm, 0.57±0.69mm, 0.96±1.31mm, and −0.92±3.12mm, respectively. For sizing, ADMIRE performs best in vitro, but in the ex vivo study delineation of the border was unclear. For detection, the custom MLSC method was able to achieve excellent discrimination between the stones and the diffuse scattering media with a constant threshold across all sets.