Ultrasonic attenuation imaging using spectral cross-correlation and the reference phantom method

Estimation of ultrasonic attenuation properties of soft tissue provides an additional diagnosis tool that may help distinguish benign from malignant tumors and detect diffuse disease. Previously introduced methods of obtaining attenuation slope estimates using medical ultrasound include spectral difference and spectral shift based methods. Recently a hybrid spectral domain method which combines the ideas of using a reference phantom with known attenuation and backscatter properties to reduce system effects and using spectral cross-correlation for frequency downshift estimation was described. The hybrid method reduces the impact of system dependent parameters including diffraction effects by normalizing the sample power spectrum by the reference power spectrum. Although the hybrid method provides improvements with regard to estimation accuracy and reduced impact of backscatter level changes, it shows resolution limitations and deficiencies in background suppression when used to generate attenuation images. Hence, it is unable to provide localized attenuation information. In this paper we provide a theoretical derivation demonstrating a linear relationship between the frequency downshift of normalized power spectrum with imaging depth, without reintroducing the transmit pulse shape to the normalized power spectrum.