3D Temperature Imaging Using Ultrasonic Backscatter Energy during Non-Uniform Tissue Heating

Background: Hyperthermia alone or in conjunction with chemotherapy and radiation is used for cancer treatment. One of its limitation is lack of detailed temperature monitoring. Ultrasound is a cheap, non-ionizing and convenient method with potential for non-invasive temperature imaging. Previously Straube and Arthur predicted monotonic changes in backscattered energy (CBE) of ultrasound with temperature. Measured CBE values from bovine liver, turkey breast, and pork muscle in 1D and 2D matched their prediction. In this study, the volumetric (3D) change in ultrasonic backscattered energy (CBE) was calibrated and used to estimate temperature during non-uniform heating. Methods: For accurate temperature measurement, a grid of thermocouples was calibrated using a NIST- traceable thermometer. 3D ultrasonic data sets were obtained by moving a 7.5 MHz linear, phased-array transducer in 0.6 mm steps in elevation. CBE was computed from a ratio of motion-compensated, envelope-detected images and a reference ultrasonic image, typically taken at 37°C. CBE curves obtained from turkey breast muscle were well matched by a linear regression that had a slope of 0.3dB/°C. The effectiveness of this value was tested on the 3D data using crossvalidation. Temperature estimation was also done during non-uniform heating. To evaluate the effects of noise, scatterer distribution, and spatial resolution on estimation errors, thermal modeling was done for non-uniform heating using finite element methods. Temperature estimation was tested during non-uniform heating in both gelatin and tissue phantoms. Specimens were heated from a central source so that the spatial temperature pattern decreased radially. Temperature images were computed from CBE maps using the appropriate CBE sensitivity. Results: Cross-validation study during uniform heating had 3D temperature estimation errors less than 0.5°C over 20 1cm3 volumes. Temperature maps for the gelatin phantoms with homogeneous scatterer distribution showed roughly concentric heating patterns. As expected, tissue exhibited a more heterogeneous heating pattern. Estimated temperature maps were validated using thermocouple readings at locations distributed throughout the specimens. Estimation errors during non-uniform heating were typically within ±1°C. Conclusion: This work, which validated the potential of CBE as a non-invasive thermometer during both uniform and non-uniform heating, was the first of its kind. It also helped in identifying some of the sources of estimation errors. 3D validation of CBE thermometry in vitro is an important step in making the transition from the laboratory to the clinical application of CBE temperature imaging for hyperthermia and other thermal therapies