Experimental Validation of the Spectral Fit Algorithm Using Tissue Mimicking Phantoms

Characterizing tissue by measuring its physical properties has potential for improving the diagnostic capabilities of medical ultrasound. One parameter that is estimated using the power spectrum of backscattered ultrasound echoes is the characteristic size or correlation length of the tissue microstructure responsible for the scattering. Unfortunately, the characteristic size measurement has traditionally been corrupted by uncertainties in the frequency-dependent attenuation along the propagation path. Recently, this problem was addressed by estimating the characteristic size and attenuation along the propagation path simultaneously using an algorithm termed the Spectral Fit (SF) algorithm. Using computer simulations to study the SF algorithm, the accuracy and precision of the attenuation estimate improved as the frequency range (largest frequency minus smallest frequency used by the algorithm when obtaining estimates) was increased. Similarly, the accuracy and precision of the scatterer size estimates were improved as the Δkaeff (largest wavenumber times scatterer radius minus smallest wavenumber times scatterer radius used by the algorithm) was increased. In this study, the simulation results were validated using a tissue mimicking phantom. The phantom had an attenuation of 0.83 dB/cm-MHz with glass beads with a radius range between 22.5 and 26.5 μm at a concentration of 47.7/mm. The SF algorithm accurately estimated the size of the glass beads and the precision of the size and attenuation estimates improved with increasing Δkaeff and frequency range respectively in the same manner as was observed in the simulation studies