Elasticity Imaging for Trauma

This research effort deals with extending the capabilities and applications of tissue elasticity imaging. Elasticity imaging is of interest, in part because soft tissues, from normal to pathological, exhibit a greater variation in elastic parameters than in acoustic parameters. We have developed a much faster alternative to the conventional cross-correlation method are interested in evaluating both 2D and 3D elasticity imaging and to investigate the use of force sensors for converting strain images into stress images. In addition, we have developed a much faster alternative to the conventional cross-correlation method, termed Analytical Phase Tracking.

This is a brief description of the Analytical Phase Tracking Method. For 1D axial displacement estimation, a sequence of RF signals as a function of slow time (applied external force function) is recorded. Via the Hilbert transform, amplitude and wrapped phase of the analytical signal are obtained along fast time. The (fast time wrapped) phase values for all RF signals are stored in a phase matrix, where row and columns represent slow and fast time, respectively. Let the analytical signal phase at a specified time t0 be φ0; our method is an efficient way of tracking φ0 across the phase matrix, while recording the time shift in the fast time location of φ0, proportional to displacement of scatterers initially at t0. In our current implementation, only phase values near 0 radians are retained, as illustrated by the phase bands in part (a), which can have bifurcation or other anomalies when the corresponding amplitude is low. Therefore, an amplitude threshold is applied to the phase matrix, giving the result in part (b). Finally, discontinuous phase bands are removed, with only connected phase bands retained, as seen in (c). Connected component labeling is then used to recognize zero phase trajectories, and slow time shifts are computed by subtracting row index at pre–compression.  The axial strain is estimated by applying 2D linear least square fitting to displacement maps.

The method was first tested with a simulated tissue-mimicking 30 mm × 15 mm phantom with a 5 mm diameter circular inclusion at its center whose stiffness was 5 and 10 times that of the surrounding tissue. Using finite element analysis (FEA) method (Comsol), axial compressions of 0.8%, 5% and 10% were applied to the phantom, and the ultrasound RF signals of the compressed medium were simulated with Field II. The DPM and strain were computed with the phase tracking method and compared to the results from FEA and standard CC. The method was then tested experimentally on a tofu phantom, using an Ultrasonix RP 4MHz transducer at a high frame rate. A time sequence of 2D RF lines for several cycles of the manual forcing function was acquired and the corresponding DPM along slow time were determined and compared with CC results.

a) Phase matrix showing wrapped phase values near 0 radians (limited in +/-0.2π)

(b) Phase matrix with normalized amplitude threshold applied

(c) Phase matrix with amplitude threshold applied and discontinuous phase curves removed