Simulation, reconstruction and effect of phase-sensitivity in ultrasonic transmission tomography

This project studies theories related to x-ray (or straight line) tomography using a synthetic image. Projections of a test object are taken at different angles before being reconstructed back into a cross-section, technically called a tomogram. Since visual inspection of the image quality, as used by other researchers, can be inaccurate due to its subjective nature, this project applies the structural similarity index as a tool for image quality evaluation. In straight line tomography, increasing the number of projections and applying a filter results in a greater index and also a better image quality. After studying straight line tomography, this work investigates finite-element modelling of ultrasonic tomography. In particular, this project investigates the relative performance of phase-sensitive and phase-insensitive receivers. These receivers are different in terms of their treatment of the signal. A phase-insensitive sensor destroys the phase information with respect to the transmitter by responding to the intensity field (or a multiple of the intensity). Computationally, this corresponds to combining a square of both real and imaginary part from each sensor element before averaging all the values into one representative value. The value is then used for the reconstruction. On the other hand, a phase-sensitive sensor retains all the amplitude and phase of the detected signal. This corresponds to averaging real and imaginary pressure from each sensor element separately before taking an absolute value. According to the study, it is found that the receiver size and the phase sensitivity affect the image quality.In general, the smaller the sensor, the better the image quality. Phase-sensitivity is also shown to greatly degrade the image when the receiver is large. Filtering, instead of improving the image quality, greatly reduces some of the detail when working with ultrasonic data. A basic ultrasonic tomography experiment is also carried out using a sample made of agar. It is found that the phase-insensitive sensor generates a better image quality for larger receivers. However, if the phase-insensitive sensor is not available, a smaller phase-sensitive receiver may be used to attain an acceptable image quality. This finding can be very useful in medical and industrial ultrasonic tomography, in order to improve or develop the imaging equipment. Furthermore, selected iterative reconstruction algorithms have been studied and implemented. They require some initial knowledge about the object as an initial estimation. It is found that the Algebraic Reconstruction Technique (ART) algorithm works slowly but results in an acceptable image quality. The ART can also be applied to both straight line and ultrasonic data without any problems. On the other hand, the Maximum Likelihood Expectation Maximisation (MLEM) generates an unnatural image, despite the quicker computation