Computer modelling of iterative technique application for tissue thermal imaging

Thermal therapies are effective methods used for cancer treatment. During thermal cancer treatment there are usually small temperature fluctuations of between 37°C and 43°C inside the human body. These fluctuations are very important to monitor accurately for quality cancer treatment, preferably with a non-invasive technique. Through-transmission ultrasonic tomography is a convenient non-invasive technique, which allows internal cross-sectional images of acoustic properties to be obtained; acoustic properties change as the temperature changes, and this allows temperature fields in tissue to be measured. Computer modeling of temperature fluctuation reconstruction using ultrasonic through-transmission tomography has been carried out. The model considers a fan-bean tomography scheme (60 sensors, 2160 rays in total) and iterative techniques for solving the inverse problem. It also assumes the 2D temperature is being reconstructed in an area of tissue of 60mm × 60mm size, and the temperature fluctuation size is ∼6.5 mm in diameter. Such heated regions can appear, for example, during thermal treatment using high intensity focused ultrasound (HIFU). Algorithms for solving the inverse problem using algebraic iterative methods (including EM, ART and SART) have been investigated. Bilinear interpolation of temperature values was used in between the nodes, in which temperature is being reconstructed. The model takes into account electronic measurement noise and assumes increasing linear dependence of sound speed on temperature, which is approximately correct for the considered temperature range and non-fatty tissue. Circular and ellipsoidal heated regions have been investigated, along with multiple contrasts. It was shown that it is possible to reconstruct temperature fluctuations with a good accuracy. A reconstruction technique was also investigated for regions of tissue which contain bones or other acoustically non-transparent objects, as these objects cause missing parts in the projection data. A special method for temperature distribution reconstruction from noisy projections with missing data was developed. The developed method reduces the number of unknowns by partial substitution of values obtained by iterative EM technique reconstruction. The accuracy of the developed method was investigated by computer modeling and compared with known expectation maximization methods. It was shown that the developed method gives higher accuracy and less distortion after reconstruction than the expectation maximization method. The developed method, therefore, may give a better opportunity to accurately monitor the thermal cancer treatment of more regions of the human body.