Task-oriented comparison of power spectral density estimation methods for quantifying acoustic attenuation in diagnostic ultrasound using a reference phantom method.

1University of Wisconsin, Madison, WI, USA.
Ultrasonic Imaging (Impact Factor: 1.16). 07/2013; 35(3):214-34. DOI: 10.1177/0161734613495524
Source: PubMed

ABSTRACT Reported here is a phantom-based comparison of methods for determining the power spectral density (PSD) of ultrasound backscattered signals. Those power spectral density values are then used to estimate parameters describing α(f), the frequency dependence of the acoustic attenuation coefficient. Phantoms were scanned with a clinical system equipped with a research interface to obtain radiofrequency echo data. Attenuation, modeled as a power law α(f)= α0 f (β), was estimated using a reference phantom method. The power spectral density was estimated using the short-time Fourier transform (STFT), Welch's periodogram, and Thomson's multitaper technique, and performance was analyzed when limiting the size of the parameter-estimation region. Errors were quantified by the bias and standard deviation of the α0 and β estimates, and by the overall power-law fit error (FE). For parameter estimation regions larger than ~34 pulse lengths (~1 cm for this experiment), an overall power-law FE of 4% was achieved with all spectral estimation methods. With smaller parameter estimation regions as in parametric image formation, the bias and standard deviation of the α0 and β estimates depended on the size of the parameter estimation region. Here, the multitaper method reduced the standard deviation of the α0 and β estimates compared with those using the other techniques. The results provide guidance for choosing methods for estimating the power spectral density in quantitative ultrasound methods.

  • [Show abstract] [Hide abstract]
    ABSTRACT: This work is a simulation-based investigation of the performance of adaptive multitaper (aMTM) windows for reducing apparent coherence in spectral analysis of ultrasound echo signals. The multitaper method may be useful when echo signal segments are limited in their axial or lateral extent. The motivation is to create high spatial resolution, low-noise parametric images of Quantitative Ultrasound (QUS) parameters derived for incoherent scattering. Pulse-echo simulations convoluted a broadband (25%100% fractional bandwidth) acoustic pulse with a one-dimensional assortment of randomly distributed scatterers. Sets of 3000 independent, simulated echo signals were computed for different bandwidths and for various concentrations of scatterers (2-20 scatteres per pulse length). The power spectral density (PSD) was estimated as the average of individual periodograms from gated segments of a subset of the independent signals. PSD estimates were computed with the Short Time Fourier Transform using low-leakage tapering functions, Welch's method with different subsegment length and shift ratios, and Thomson's multitaper method. An adaptive time bandwidth selection criterion was designed to estimate PSD-derived parameters such as the backscatter coefficient using the multitaper method. The mean squared error (MSE) of PSD estimates was computed when reducing the window size from 1 to 50 pulse lengths axially and 1 to 50 averaged realizations laterally. For a particular MSE value, the window size leading to equal contributions of the bias and coherent noise was determined, and the diagonal of this window (Dw) was used as a criterion for comparison among PSD estimation methods. The adaptive multitaper method led to 77% and 13% reductions in Dw compared to that of the Short Time Fourier Transform (regardless of the windowing function) and Welch's method, respectively. These values did not vary significantly with different pulse bandwidths or scatterer densities above 10 scattere- s per pulse lengths. The adaptive multitaper method successfully reduced bias and coherent noise compared to values for other methods, showing its advantage for spectral analysis of incoherent backscattered signals.
    2013 IEEE International Ultrasonics Symposium (IUS); 07/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Quantifying features related to acoustic scatterer periodicity can provide useful information to monitor tissue structural changes, but their detection is hindered by apparent coherence from random scatterers. This work compares the use a multitaper Generalized Spectrum (mtGS) to single-taper and time-average approaches (stGS and taGS, respectively) and to the Singular Spectrum Analysis (SSA) for detecting periodicity in backscattered echo signals when reducing the size of the parameter estimation region. A phantom with diffuse scatterers and an array of 0.1mm-diameter nylon fibers 0.4mm apart was scanned with a Siemens S2000 system using a linear array transducer. Radiofrequency (RF) echo signals from the fiber plane were obtained and Generalized Spectrum (GS) estimates were made either by stGS, taGS or mtGS with Discrete Prolate Spheroidal Sequences. Spectral components corresponding to periodic structures were identified by peaks in the GS Collapsed Average. SSA was implemented by obtaining eigenvalues and eigenvectors of the autocovariance matrix of signal segments. The periodic components of envelope signals were reconstructed using pairs of eigenvectors with similar eigenvalues. The frequency of the periodic component was estimated from the maximum value of its power spectrum. Histograms of frequency components detected by each method were constructed. The conspicuity of the 1.9MHz peak (corresponding to the fiber spacing) was measured as the size of the parameter estimation region was reduced axially and laterally from 20 to 2 correlation lengths. The mtGS improves detection of the relevant frequency components (1.9MHz and its harmonic) compared to stGS, taGSm and SSA by increasing their conspicuity over spurious components. This method also provided the minimum parameter estimation region size (8 pulse lengths axially, 6 uncorrelated scanlines laterally) viable for detection of periodic features.
    2013 IEEE International Ultrasonics Symposium (IUS); 07/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Acoustic form factors have been used to model the frequency dependence of acoustic scattering in phantoms and tissues. This work demonstrates that a broad range of scatterer sizes, individually well represented by Faran theory or a Gaussian form factor, is not accurately described by a single effective scatterer from either of these models. Contributions from a distribution of discrete scatterer sizes for two different form factor functions (Gaussian form factors and scattering functions from Faran's theory) were calculated and linearly combined. Composite form factors created from Gaussian distributions of scatterer sizes centered at 50 µm with standard deviations of up to σ = 40 µm were fit to each scattering model between 2 and 12 MHz. Scatterer distributions were generated using one of two assumptions: the number density of the scatterer diameter distribution was Gaussian distributed, or the volume fraction of each scatterer diameter in the distribution was Gaussian distributed. Each simulated form factor was fit to a single-diameter form factor model for Gaussian and exponential form factors. The mean-squared error (MSE) between the composite simulated data and the best-fit single-diameter model was smaller with an exponential form factor model, compared with a Gaussian model, for distributions with standard deviations larger than 30% of the centroid value. In addition, exponential models were shown to have better ability to distinguish between Faran scattering model-based distributions with varying center diameters than the Gaussian form factor model. The evidence suggests that when little is known about the scattering medium, an exponential scattering model provides a better first approximation to the scattering correlation function for a broad distribution of spherically symmetric scatterers than when a Gaussian form factor model is assumed.
    Ultrasonic Imaging 05/2014; DOI:10.1177/0161734614534399 · 1.16 Impact Factor