Shear Wave Imaging of the Breast

n the February 2012 issue of the Journal of Ultrasound in Medicine, Bai et al1 presented their results on the evaluation of solid breast masses with quantitative acoustic radiation force impulse (ARFI) imaging. In their study, they found that 63.4% of breast cancers had a result of “X.XX m/s,” indicating that a measurement result could not be obtained. Quantitative ARFI imaging (Virtual Touch tissue quantification; Siemens Medical Solutions, Mountain View, CA) uses algorithms that include a test of the confidence interval of the shear velocity estimation. The algorithm rejects the data if the confidence interval is low (<0.8) due to several quality factors, including shear wave amplitude and noise. If the data are rejected, the system gives a speed of the shear wave (Vs) as X.XX m/s. We have seen similar results in our investigations. There are several reasons why an X.XX m/s result can occur, including improper data collection (tissue movement and probe movement), the shear wave speed exceeding the maximum measurable range of 8.4 m/s (limit of the ARFI technique in this case), or a poor-quality shear wave. With breast imaging, motion is easier to eliminate than in abdominal imaging, in which respiratory motion and/or vascular motion is a greater problem. In our experience, motion is easily controlled in breast elastography, and it has not been a problem. In these cases, multiple repeated measurements in the same lesion return an X.XX-m/s result. Although possible, it is unlikely that all of these breast cancers have a shear wave speed exceeding 8.4 m/s. We are able to capture the raw data signal in our research system from clinical cases and process the shear wave data off line. Figure 1 illustrates the principle of quantitative ARFI imaging. A indicates the excitation pulse; B is the region of interest where shear waves are detected; C is a representation of the shear wave displacement profiles at distance intervals away from the excitation pulse; and D indicates the linear regression analysis used to estimate the shear wave velocity. In Figure 2, the typical shear wave pattern from a fatty area in a patient’s breast is presented. Each individual curve taken at different distances from the push pulse has minimal noise. The time to peak for each curve is used to calculate the shear wave speed. Figure 3A is a similar display from an invasive ductal cancer. Note that the shear wave amplitude is low and noisy from within the cancer, but a smooth shear wave response is seen in the surrounding normal tissue. Further evaluation is required to understand the cause of this result and will not be addressed here. In this case, a shear wave speed from within the cancer can not be accurately measured. The ARFI algorithms reject these data as poor, therefore giving a result of X.XX m/s. Invited paper I