Hyperthermia is a cancer treatment in which tumors are elevated to temperatures toxic to cells (41–45° C).3 It has been used to augment the effects of chemotherapy and radiation. Unfortunately, lack of detailed temperature information to guide thermal therapy limits its broader use.1 A noninvasive, clinically useful method is needed to measure 3D temperature distributions to within 0.5° C in 1cm3 volumes.
Noninvasive temperature-estimation methods based on magnetic resonance imaging (MRI) and ultrasonic images have received the most attention.2,7 The required accuracy and spatial resolution can probably be achieved with MRI, but it is expensive and may be difficult to use in conjunction with some thermal therapies. On the other hand, ultrasound is inexpensive, nonionizing, portable, and convenient. It also has relatively simple signal-processing requirements. These qualities make it an attractive candidate for temperature estimation. Ultrasonic tissue properties affected by temperature include speed of sound, attenuation, and backscattered energy.
Backscattered Energy in Motion-Compensated Images
To find an ultrasonic parameter that changes monotonically with temperature, we modeled the backscattered energy from individual scatterers.
4 According to our model, the backscattered energy could change by a factor of two to three over the temperature range from 37 to 50° C.
5 Change in backscattered energy (CBE) was modeled assuming that the scattering potential of the volume was proportional to the scattering cross-section of sub-wavelength scatterers. We predicted with this model that the change in backscattered energy could increase or decrease depending on the type of inhomogeneity causing the scattering.
In addition, to measure changes in backscattered energy in the hyperthermia treatment range, conventional phased-array images were taken as a function of temperature.6,8 Tissue was heated in an insulated tank filled with deionized, degassed water. The temperature ranged from 37 to 45° C for measurements on nude mice in vivo and up to 50° C for specimens of bovine liver, turkey breast, and pork muscle in vitrio. The nude mouse image in Figure 1 highlights a region of interest including the leg, which had been implanted with an HT29 human colon cancer tumor.
Figure 1. Ultrasound image of a nude mouse in vivo at 37° C produced with a Terason 2000 system with a 128-element 7MHz array.
To determine CBE accurately, the same regions must be compared as temperature changes. Because the speed of sound changes with temperature, even in vitrio images appear to move. Of course, real motion is also present in vivo. To estimate feature displacement, we maximized cross-correlation between radio-frequency images at adjacent temperatures using a combination of the optimization and image re-sampling functions available in Matlab™. This eliminated dependence on the spatial sampling period of the image.
8
After motion compensation, CBE was calculated over the measured temperature range. Envelopes of motion-compensated image regions were found with the Hilbert transform, then smoothed with a 3×3 running average filter. Values were squared to determine the backscattered energy at each pixel.
CBE measurement, prediction, and simulation
Figure
2 shows CBE over a region of interest in a specimen of bovine liver relative to the energy at 37° C. The backscattered energy increased (red) with temperature for some regions and decreased (blue) in others. Figure
3 shows the means over the pixels with increasing energy (red) and over the pixels with decreasing energy (blue) in eight regions of interest in a turkey breast specimen. From 37 to 50° C, the increase is about 4dB and the decrease about 3dB. Our single-scatterer model predicted this CBE range, as shown in Figure
4.
4
Figure 2. Change in backscattered energy (CBE) in ultrasound images of bovine liver from 37 to 50° C after compensation for apparent motion. All images were referred, pixel-by-pixel, to the energy in the reference image at 37° C. Each colorbar is in dB.
Figure 3. Means of measured positive (red) and negative (blue) changes in backscattered energy in eight regions of interest in a specimen of turkey breast.
Figure 4. Predicted CBE for single, sub-wavelength lipid, and aqueous scatterers in an aqueous medium from our previous study.5
To extend the single-scatterer model, we simulated images from different populations of multiple scatterer types.
9 Images from collections of thousands of discrete, temperature-dependent, sub-wavelength scatterers have allowed us to study the effects of scatterer type, region size, and noise on CBE. Addition of noise to the simulations yields results very similar to our measured CBE values (see Figure
3) and suggests that estimated temperatures with errors of 0.5°C are possible in 1cm
3 volumes.
Conclusions
Changes in backscattered energy have been monotonic in predictions, simulations, and in both in vitro and in vivo measurements. This behavior enables temperature estimation from ultrasonic images. Successful CBE temperature estimation based on 3D tracking and compensation for image feature motion could eventually serve as the foundation for the generation of 3D temperature maps in soft tissue as a noninvasive, convenient, and low-cost way to regulate temperature during hyperthermia treatment of cancer.
R. Martin Arthur, Jason W. Trobaugh, William L. Straube
Department of Electrical and Systems Engineering, USA
Washington University in St. Louis
St. Louis, MO
Eduardo G. Moros
University of Arkansas for Medical Sciences
Department of Radiation Oncology, USA
R. Martin Arthur is the Newton R. and Louisa G. Wilson Professor of Engineering at Washington University. His current research interests include synthetic-focus methods for image improvement and tissue characterization using medical ultrasound
Jason W. Trobaugh is a research instructor in the School of Medicine and a research associate in the School of Engineering and Applied Science at Washington University. His research interests are in the fields of model-based image analysis, probabilistic image models, and treatment guidance applications for ultrasonic imaging.
William L. Straube is a research associate professor at the Washington University School of Medicine. His interests include developing and applying new thermal therapies.
Eduardo G. Moros is director of the Division of Radiation Physics and Informatics at the University of Arkansas. His main interest is the development of imaging/therapy technology to advance the treatment of cancer with heat and ionizing radiation.
References:
3.
J. Gellermann, W. Wlodarczyk, H. Ganter, J. Nadobny, H. Fahling, M. Seebass, R. Felix, P. Wust, A practical approach to thermography in a hyperthermia/magnetic resonance hybrid system: Validation in a heterogeneous phantom,
Int. J. Radiat. Oncol. Biol. Phys.
61, no. 1, pp. 267-277, 2005.
