King's College London

Research portal

Design of a Microwave Tomography System for Medical Imaging Applications

Student thesis: Doctoral ThesisDoctor of Philosophy

Microwave tomography (MT) is an emerging imaging modality which aims to recover the dielectric profile of a domain by solving an inverse problem. This is a challenging problem that requires sophisticated algorithms as well as hardware design. This thesis presents a simple and low cost design of a MT system that can operate in the 1-3 GHz frequency range. The hardware prototype of the system is developed from concept to physical realization and is validated against numerical and experimental studies using an in-house inversion algorithm. As with most experimental MT systems presented in the literature, this thesis focuses on cylindrical setups that can be imaged efficiently with a two-dimensional (2-D) inversion algorithm.
Using an antenna that can operate efficiently in the desired MT frequency spectrum is essential for any MT system. To this end, the thesis studies several antenna designs and evaluates their performance by calculating return loss and transmission levels in the desired frequency range. Based on this analysis, we select a custom-made printed monopole antenna with very small size which can operate efficiently across the selected frequency range when immersed in various materials that are used as coupling liquids in microwave tomographic systems.
The selection of coupling medium is quite vital in the design of a microwave imaging system, which is subject to various signals that obscure the response from the object to be imaged. In particular, multipath signal propagation and surface waves along with other degrading factors such as noise, coupling etc., pose significant challenges to data integrity. We address this issue by studying the performance of the selected antenna (stand-alone and as an array) in various coupling liquids, such as water mixed with glycerine or corn syrup. The aim of this study is to ensure that the sensitivity of our MT system is sufficient to detect weak target responses in a practical measurement with a standard vector network analyzer (VNA), while at the same time effects such as antenna coupling and multi-path propagation are minimised.
The thesis also presents a comparison of two possible MT setups: non-immersed con-figuration, where the array of antennas operates in free space but very close to the imaging chamber, and an eight-element antenna array immersed inside different coupling liquids. The array performance in the aforementioned configurations is also experimentally assessed, by acquiring data with a two-port vector network analyser (VNA). In terms of the reconstruction scenarios, we focus on two cases that are studied numerically and experimentally: a target inside a cylinder filled with coupling medium, and the target inside a cylinder filled with a low-loss liquid, surrounded by the coupling liquid where the antennas are immersed.
Comparison of our experiments to numerical data suggests that measurements are very sensitive to errors such as cable movements or imprecise spatial positioning of antennas relative to the imaging chamber. To circumvent these issues, we synthesise the array by acquiring data in a bi-static configuration using a robust, mechanically calibrated system which can guarantee accurate antenna positioning. Comparing the measurement data from the bi-static configuration with CST simulation results lead to a much better agreement. This suggests that a significant source of measurement error can be introduced if the multi-static system is not designed with a lot of care on cables positioning.
An in-house inversion algorithm is applied to the acquired data to validate our system’s ability to reconstruct cylindrical targets. The cylindrical target is reconstructed successfully using the inversion algorithm with both experimental and simulation data for both imaging scenarios of one and two-layer phantoms. The system is among world’s first experimental imaging systems that can reconstruct targets successfully in the wide frequency range of 1-3 GHz.
Original languageEnglish
Awarding Institution
Award date2018


View graph of relations

© 2018 King's College London | Strand | London WC2R 2LS | England | United Kingdom | Tel +44 (0)20 7836 5454