Task-oriented comparison of power spectral density estimation methods for quantifying acoustic attenuation in...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Acoustics

Background:

• Quantitative ultrasound (QUS) methods rely on accurate estimation of acoustic attenuation.
• Acoustic attenuation is frequency-dependent and often modeled using a power law: α(f) = α0 f^β.
• Accurate estimation of attenuation parameters (α0 and β) requires reliable power spectral density (PSD) estimation from backscattered ultrasound signals.

Purpose of the Study:

• To compare the performance of different power spectral density (PSD) estimation methods for quantitative ultrasound.
• To evaluate how parameter estimation region size affects the accuracy of acoustic attenuation coefficient (α(f)) estimation.
• To provide guidance on selecting optimal spectral estimation techniques for QUS applications.

Main Methods:

• A phantom-based study was conducted using radiofrequency echo data from a clinical ultrasound system.
• Power spectral density (PSD) was estimated using the short-time Fourier transform (STFT), Welch's periodogram, and Thomson's multitaper techniques.
• Attenuation parameters (α0 and β) were estimated using a reference phantom method, with errors quantified by bias, standard deviation, and overall power-law fit error (FE).

Main Results:

• All spectral estimation methods achieved a low overall power-law fit error (FE) of 4% for parameter estimation regions larger than approximately 34 pulse lengths.
• For smaller estimation regions, typical in parametric imaging, the bias and standard deviation of α0 and β estimates were dependent on region size.
• The Thomson multitaper method demonstrated a reduced standard deviation for α0 and β estimates compared to STFT and Welch's method in smaller regions.

Conclusions:

• The choice of spectral estimation method significantly impacts the accuracy of attenuation parameter estimation, especially in limited data regions.
• The Thomson multitaper technique offers advantages in reducing variability for attenuation estimation in QUS, particularly for parametric imaging.
• These findings offer practical guidance for researchers and clinicians in selecting appropriate methods for power spectral density estimation in quantitative ultrasound.