Limits imposed by tissue on SVD-Wigner filtering to obtain the intrinsic frequency variation of ultrasonic echo waveforms

The attenuation slope with frequency from the frequency decrease along the ultrasonic echo signal is evaluated in tissue. When the local frequency from the reflected waveform is evaluated, many data segments yield unreliable values because of interference between overlapping echo pulses. The authors advance an analysis based on the singular-value decomposition (SVD) of the Wigner distribution of the signal, which provides a criterion to filter out data segments corrupted by interference. When applied to clinical data, this SVD-Wigner filtering process reduces the scatter in the local frequency estimate sufficiently to produce a reliable estimate of the frequency slope along the signal with about one third of the data needed by conventional short-time Fourier techniques. The improvement is not larger because these clinical data were taken on a scanner with a large sample volume, so that most data segments are perturbed by interference artifacts. Simulations show that in a more tightly focused system, with fewer scatterers contributing to the echoes, the reduction in required data can be appreciable