Quantitative Ultrasound Comparison of MAT and 4T1 Mammary Tumors in Mice and Rats Across Multiple Imaging Systems

Quantitative ultrasound estimates based on the backscatter coefficient (BSC) have demonstrated the ability to detect and classify disease beyond the capabilities of traditional B-mode imaging. Quantitative ultrasound has been able to differentiate live from apoptotic cells or oncotic cells in vitro with high-frequency ultrasound,1,2 monitor changes in tissue microstructure due to high-intensity focused ultrasound at high frequencies in a preclinical model,3 and monitor changes due to chemotherapy clinically at 6 MHz.4 Tumors have been characterized and differentiated using quantitative ultrasound at high frequencies in preclinical models of thyroid cancer,5 at high frequencies in clinical applications for ocular tumors,6,7 and at clinical frequencies in human patients for breast tumors8 and prostate cancers.9,10 Also, the BSC has been proven valuable for diagnosing fatty liver disease clinically11 and for monitoring physiologic and pharmacologic responses in renal function at clinical frequencies in an animal model.12,13 One of the BSC features that is often cited as an advantage is its system independence; ie, the same BSC value for a given tissue can be estimated from any imaging platform. Previous studies have demonstrated system independence of BSC estimates in well-characterized phantoms.14–17 These investigations applied single-element imaging systems measuring backscatter levels for strong glass bead scatterer phantoms (1–12 MHz),14,15 single-element imaging systems for probing weakly scattering agar-in-agar phantoms (1–13 MHz),16 and clinical array imaging systems for estimating BSC for phantoms with glass bead scatterers (1–15 MHz).17 Parametric estimates based on the BSC have demonstrated the ability to differentiate different types of orthotopic mouse tumors ex vivo (16–24 MHz),18 to identify regions of micrometastases within resected lymph nodes clinically (26.5 MHz center),19 and to identify dominant sources of backscattering in the renal cortex (frequencies covering 2.5–15 MHz).12,13,20,21 The frequency ranges in these studies cover the range being investigated in this study: namely, from the clinical frequencies to the lower end of the small-animal pre-clinical imaging frequencies, covering, for example, 1 to 21 MHz. This study addresses the need for demonstrating that in vivo interlaboratory measurements can be used to distinguish different types of tumors, while being able to obtain consistent BSC results from all systems. Previous studies22,23 examined BSC agreement across single-element and clinical imaging systems for an in vivo rodent fibroadenoma model over a frequency range of 1 to 14 MHz. The heterogeneity of the benign fibroadenomas provided a near worst-case scenario to evaluate quantitative ultrasound performance. The study presented herein seeks to expand on this work using clinical and preclinical array imaging systems to test their capability to detect known tissue differences while maintaining good agreement in BSC values across all systems. Mouse (4T1) and rat (MAT) mammary tumor models were used in a comparative experiment with 3 array-based ultrasound imaging systems (Siemens, Ultrasonix, and VisualSonics) and 5 different transducers covering a range of frequency bandwidths. Backscatter coefficient agreement between systems was assessed, as well as the ability to differentiate the two tumor types.