Figure 1.

Block diagram of the Pulse Wave Imaging (PWI) method on a normal human aorta in vivo. (A) A sequence of RF frames is acquired at the minimum depth required to visualize both walls of the artery of interest. (B) A 1D normalized cross-correlation-based motion estimation method (Luo and Konofagou 2010) is used on the RF signals to compute the inter-frame axial (i.e. parallel to the ultrasound beams) displacements in the arterial walls. The displacement amplitudes are normalized by multiplying by the frame rate. The arrival of the pulse wave induces positive (i.e. towards the transducer) displacements in the anterior wall and negative (i.e. away from the transducer) displacements in the posterior wall. The white arrows indicate the propagation of the wavefront along the anterior wall. (C) The anterior wall is manually segmented (red dotted line) in the first frame of the acquisition sequence, generating a wall trace that specifies the depth of the wall at each scan line (i.e. one value per scan line). This trace is automatically updated in subsequent frames based on the estimated displacements. (D) The wall motion is spatiotemporally mapped by plotting the displacement at each point along the wall trace over time. (E) The normalized displacement waveform at three scan line positions along the anterior wall, corresponding to the light gray, dark gray, and black lines in (D), are shown. The time point corresponding to a characteristic tracking feature (e.g. the 50% upstroke as indicated by the blue dots) is automatically detected in the waveform at each position. (F) Linear regression is performed on the spatio-temporal variation of the characteristic time points along the imaged segment to obtain the slope as the PWV and the coefficient of determination r² as an index of propagation uniformity.