Experimental validation of the structure factor model on tissue-mimicking phantoms with high scatterer concentrations

Quantitative ultrasound technique is based on a frequency-based analysis of the signals backscattered from biological tissues. This technique aims to estimate the size and concentration of scatterers in order to diagnose and monitor diseases, such as cancer. The Gaussian Model (GM) and Faran Model (FM) have been used for many years but are limited to dilute scattering medium, whereas the scatterers can be densely packed (for example the cells in cancer). A model adapted to dense medium is the Structure Factor model (SFM) used in blood characterization. However, the most often used SFM version is the Percus Yevick model (PYM) using the low frequency limit of the structure factor called the Percus Yevick packing factor. The aim of this work is to compare the aforementioned scattering models with measured backscatter coefficients (BSCs) on tissue-mimicking phantoms. The tissue-mimicking phantoms consisted of polyamide microspheres (mean radius 6 μm) in water suspension. The phantoms had identical scatterer sizes with an impedance contrast of 58% (versus 42% in cell nuclei) but have different scatterer volume fractions ranging from 1 to 25%. Ultrasonic backscatter measurements were made for frequencies from 6 MHz to 22 MHz. Good agreement was shown between the measured frequency dependent BSCs and those predicted with the SFM. Figure 3 shows the measured BSC amplitude in the frequency bandwidth from 6 to 15 MHz versus the scatterer concentration. Also plotted are the theoretical BSC amplitude computed with the FM, PYM and SFM. Excellent agreement was obtained at a low volume fraction of 1% and 2.5% for all models. The FM (and the PYM, respectively) overestimated the BSC amplitude for volume fraction >;5% (and for larger volume fraction >;12.5%). Most of tissues, such as cancer, need to be characterized as dense scattering media. For large scatterer volume fraction, the experimental BSCs agreed with predictions using the SFM in frequency dependence and s- attering magnitude.