Design of wideband microwave frontend for microwave-based imaging systems

Abstract

Microwave imaging has been investigated in the last few years as an attractive complement to current diagnostic tools for medical applications due to its low-cost, portability and non-ionization radiation. Research to verify the feasibility of microwave imaging systems for medical applications is conducted using a Vector Network Analyser (VNA) as the microwave transceiver. The VNA is usually bulky and expensive, and thus prevents microwave imaging systems from being low-cost and portable. A necessary condition to turn microwave imaging into a mass screening diagnostic tool is to replace the VNA with a low-cost portable unit that can characterise, generate, transmit and receive signals, across a wideband with large dynamic range and stability. Tissues in the human body are lossy at microwave frequencies, and hence microwave signals undergo high attenuation when penetrating the human tissues during the process of imaging. Using a wider frequency spectrum provides better resolution, but low frequencies penetrate further into the body. As a trade-off between the required signal penetration and image resolution, microwave frequencies within the band 0.5-4 GHz have been used in many medical applications, such as head, torso, and breast imaging. Thus, the desired generic microwave transceiver for microwave-based medical imaging should cover this wide frequency band. This thesis proposes two versions of a reconfigurable low-cost portable broadband RF frontend medical imaging systems based on Software Defined Radio (SDR) and in doing so makes four contributions to the field of microwave imaging systems. The first contribution is the design of a low-cost reconfigurable microwave transceiver based on software defined radar (SDRadar). A RF broadband circulator used to separate the transmitted and received signal and a virtual ultra-wideband (UWB) time domain pulse is generated by coherently adding multiple frequency spectrums together. To verify the proposed SDRadar system for medical imaging, experiments were conducted using a circular scanning system and directional antenna. An image reconstruction algorithm used to generate and verify the images of the target embedded in a phantom developed with liquid emulating the average properties of different human tissues using the measurement data. The system successfully detects and localise small targets at different locations in the phantom. The above broadband circulator, however, limited the isolation between the transmitted and received signal, and the system lacked a calibration process. Hence, the proposed system is unable to image more complex human tissues. The second contribution of the thesis is the design of Vector Network Analyzer (VNA) by using the above SDR called Software defined VNA (SDVNA) with a highly directive broadband directional coupler (0.5-4 GHz). However, using a directional coupler with a single-receiver single-transmitter the conventional open/short/match/thru calibration technique is not applicable, and thus, it is re-developed to take into account the SDVNA system’s architecture. The proposed SDVNA is capable of covering the band from 0.8 –3.8 GHz with a dynamic range of 80dB. The performance is verified by showing that the S-parameters of 1-port (antenna) and 2-ports (filter) as devices under test have close agreement to results from a commercial VNA. The system was further verified by performing microwave imaging of a head phantom with realistic permittivity and conductivity. The system was able to detect the location of an embedded bleeding in the realistic head phantom, with comparable quality to a commercial VNA, and significantly improved image compared with the previous circulator-based design. The third contribution of the thesis is the investigation of measurement accuracy of the proposed SDVNA and SDRadar for medical imaging. Measurement variation and repeatability is characterized and analysed. Measurement accuracy is essential for accurate image reconstruction. The proposed SDVNA system provides the measurement repeatability within 95% of a commercial VNA, enabling the system to be developed in the future as a low-cost portable system to be used in any clinic to detect a strokes in the brain. The fourth contribution of the thesis is the complete development of the low-cost portable multistatic head imaging system by using the above SDVNA and a helmet with array of eight antennas. In order to overcome the errors caused by antenna differences and antenna misalignment, a new calibration technique for arrays of the antennas proposed. Three known calibration standard based on free space/oil/water is developed and used effectively to eliminate the systematic error due to antenna differences and misalignment. Through the experimental analysis it was shown that the use of the proposed calibration techniques improves the location and detection accuracy of the system. The proposed low-cost portable system has a complete built in calibration techniques and auto measurement system to scan the complete head of the patient with strokes. This thesis describes the complete development of a low-cost portable medical imaging system for applications which can be used by medical professionals as a complementary imaging system to detect and localised abnormalities in human tissues.