D. Basu, R. M. Arthur, J. W. Trobaugh, Y. Guo1, and W. L. Straube, "3D Temperature Imaging using Change in Ultrasonic Backscattered Energy: Estimation of Temperature during Non-Uniform Heating", Ultrasonic Imaging, vol. 31, pp. 75-77, 2009.

Abstract

Hyperthermia is used, often along with chemotherapy and radiation, to treat cancer. It is important for clinical application of hyperthermia to raise the temperature of cancerous tissue to therapeutic levels (40-44^oC) while maintaining normal tissue at body temperature. Treatment quality and efficacy could be improved by obtaining a complete temperature profile of a region of interest. Ultrasound is a cheap, non-ionizing and convenient method to target diseased tissue and monitor temperature. Previously we predicted monotonic changes in backscattered energy (CBE) of ultrasound with temperature for sub-wavelength scatterers [1]. This theoretical model was based on the relative change in the backscatter coefficient at a given temperature to values at a reference temperature. Measured CBE values from bovine liver, turkey breast, and pork muscle in 1-D and 2-D matched our predictions [2], as did simulation results for populations of randomly distributed scatterers [3]. For clinical application of temperature imaging with CBE, our aim is to estimate temperature over tissue volumes of about 1cm³ with 0.5^oC accuracy. The objective of this study was for the first time to estimate temperature in 2D during non-uniform heating in both gelatin phantoms and turkey breast tissue using CBE based on calibration in 3D from uniform heating experiments.

For the studies reported here, CBE was calibrated and temperature estimated in both graphite-in-gelatin phantoms and specimens of abattoir turkey breast muscle. The graphite-in-gelatin phantoms were prepared following the Madsen ’78 recipe and provided a more nearly homogeneous medium than tissue. Specimens were uniformly heated from 37-45^oC. 2-D ultrasound scans were obtained every 0.6mm in the elevation direction to form 3D datasets. Temperature in tissue was measured using a thermocouple grid. Uniform heating for calibration was obtained by placing specimens in a water bath heated with a circulating heater under computer control. Ultrasonic images were acquired when the thermocouple readings were within 0.3^oC of each other. CBE was computed over 1cm³ regions by taking the ratio (pixel by pixel) of the energy in the B-mode ultrasonic image at a temperature of interest to the energy in the reference image at 37^oC. CBE curves versus temperature were well characterized by a straight line. The calibration curve for the phantoms had a slope of 0.08±0.01 dB / ^oC. Turkey breast yielded a slope of 0.29±0.03 dB / ^oC.

Temperature was estimated prospectively during non-uniform heating in both graphite-in-gelatin and turkey breast specimens. For non-uniform heating, silicon tubing was inserted through the middle of a specimen and water heated to 65^oC was passed through it. Specimens were placed in air and B-mode images were taken through gel coupling every 30s after the injection of heated water. A reference scan was taken before the beginning of the heating. The temperature in the tissue was monitored using 2 thermocouples at a distance of 1 cm and 2 cm from the tube. CBE images were obtained by moving a window of 5x5 mm over each pixel and assigning the average CBE value to that pixel. Temperature images, which are the first based on CBE, were then derived from the CBE images using the appropriate calibration curve. Both the tissue and the phantom showed a radial heating pattern, with the temperature decreasing with distance from the heated tube. Temperature maps for the phantom showed nearly concentric heating patterns, presumably due to the uniform distribution of scatterers. Tissue exhibited a heterogeneous heating pattern, likely due to the inhomogeneities in the tissue structure. Incomplete coupling of the hot water tube to the specimens and noise in the images due to gel coupling are other possibilities that affected the estimated heating patterns. The temperatures at the distance of 1 cm and 2 cm were within about 1^oC of the thermocouple readings for phantom and about 2^oC for the tissue studies. Because the thermocouples were outside the field of view of the ultrasound beam, only approximate comparison of CBE estimates and thermocouple readings was possible with the present arrangement. These studies continue to suggest the potential of CBE as a non-invasive thermometer for hyperthermia treatment.

1) WL Straube and RM Arthur, UMB, 20:915-922, 1994.

2) RM Arthur et al., IEEE Trans on UFFC, 52:1644-1652, 2005.

3) JW Trobaugh et al,,UMB, 34:289–298, 2008.

Support: R21-CA90531, R01-CA107558 and the Wilkinson Trust at Washington University, St. Louis.