Frequency-based microwave medical imaging techniques

Abstract

Medical imaging is an important technique in diagnosis and monitoring of a variety of diseases by providing a visual representation of inner organs of the human body. Although existing imaging technologies like Magnetic Resonance Imaging (MRI) and Computational Tomography (CT) can provide high resolution and accurate images, they are still expensive, bulky and time-consuming; and hence, they are not accessible in rural, remote health centres and emergency units such as ambulances. This lack of sufficient imaging tools causes millions of death each year; and therefore, a low-cost, portable, non-invasive and fast imaging device that can be used as a diagnostic and monitoring tool is deemed beneficial. In this thesis, a study on the development and use of microwave imaging systems for medical applications has been carried out. The study focuses on the development of an imaging algorithm to create images of the human body interior for medical applications. In addition, different techniques to tackle the challenges in microwave medical imaging algorithms are proposed. A fast frequency-based imaging algorithm is proposed in this thesis to mitigate problems associated with time-domain approaches. This method applies Bessel functions in frequency domain for fast image processing that maps regions of high dielectric contrasts of the imaged domain in a two-dimensional plane. In this algorithm, the Nyquist concept is adopted to reduce the number of sampling frequency utilising limited number of antennas to lower the computational complexity effectively whilst attaining correct detections. Moreover, a hybrid clutter removal technique to eliminate the effect of strong reflections from the skin on the collected signals in microwave medical imaging systems is developed as a prepossessing technique to the imaging algorithm. As the efficacy of the imaging algorithm relies on a priori information on the permittivity and boundary of the imaging domain, an antenna-specific permittivity estimation method and a boundary estimation technique by using antenna resonant frequency shifts are proposed to improve the detection accuracy and image quality. In addition, the concept of virtual antenna array to increase the effective number of antenna elements is proposed in this thesis. The performance of the imaging algorithm and the other proposed techniques are tested via full-wave electromagnetic simulations and experiments on phantoms and healthy human trials, using the head and torso imaging systems developed in the Microwave Imaging Group of the University of Queensland.