Creation and characterization of an ultrasound and CT phantom for noninvasive ultrasound thermometry calibration.

Ultrasound thermometry provides noninvasive 2-D temperature monitoring, and in this paper, we have investigated the use of computed tomography (CT) radiodensity to characterize tissues to improve the accuracy of ultrasound thermometry. Agarose-based tissue-mimicking phantoms were created with glyceryl trioleate (a fat-mimicking material) concentration of 0%, 10%, 20%, 30%, 40%, and 50%. The speed of sound (SOS) of the phantoms was measured over a temperature range of 22.1-41.1 °C. CT images of the phantoms were acquired by a clinical dedicated breast CT scanner, followed by calculation of the Hounsfield units (HU). The phantom was heated with a therapeutic acoustic pulse (1.54 MHz), while RF data were acquired with a 10-MHz linear-array transducer. Two-dimensional speckle tracking was used to calculate the thermal strain offline. The tissue-dependent thermal strain parameter required for ultrasound thermometry was analyzed and correlated with CT radiodensity, followed by the validation of the temperature prediction. Results showed that the change in SOS with the temperature increase was opposite in sign between the 0%-10% and 20%-50% trioleate phantoms. The inverse of the tissue-dependent thermal strain parameter of the phantoms was correlated with the CT radiodensity (R(2) = 0.99). A blinded ultrasound thermometry study on phantoms with a trioleate range of 5%-35% demonstrated the capability to estimate the tissue-dependent thermal strain parameter and estimate temperature with error less than ~1 °C. In conclusion, CT radiodensity may provide a method for improving ultrasound thermometry in heterogeneous tissues.