Investigation and validation of FDTD weighting function modelling for microwave radiometric temperature measurement
Microwave radiometry can provide a non-invasive, non-destructive and inherently safe method of temperature measurement suitable for a range of medical and industrial applications. The measured radiometric signal is formed by a convolution of the actual material temperature distribution with a coupling spatial response, or weighting function, over the viewed volume of material. The form of this weighting function depends on both the electromagnetic coupling structure (either antenna or cavity) and on the geometry and dielectric properties of the material. Through reciprocity, the weighting function can be found by computation or measurement of the power dissipation distribution (also known as the specific absorption rate (SAR)) when the coupling structure is actively excited. Knowledge of the weighting function is used to interpret the measured radiometric temperature. Chapter 1 introduces the method of microwave radiometry, its range of applications and considers the key features of weighting function determination. The suitability and validity of finite difference time domain (FDTD) SAR and weighting function modelling was investigated for the largely travelling - wave fields appropriate to surface contact antennas. An FDTD simulator, the Basic Electromagnetic Simulation Tool , was used to computationally model a range of antenna configurations that could then be compared directly with experimental results. Chapter 2 introduces several numerical techniques and justifies the choice of FDTD modelling. An introduction to the theory of the FDTD technique and a description of the BEST software is also given. Simulations of systems where electromagnetic field distributions are known (or can be determined experimentally) allowed the direct comparison of simulation results with theoretical predictions. Chapters 3 and 4 consider various validation examples; a monopole radiator above ground plane and TEOl waveguide in chapter 3, experimental field determination in lossy dielectrics using the non-resonant perturbation method in chapter 4. In all cases considered, simulation and experiment agree within a reasonable magnitude of error. With the successful validation of its microwave modeling capabilities, the BEST program was then used to predict the weighting functions expected for practical radiometer antennas for microwave temperature measurement. Of primary importance are the variations of the effective coupling distance into the viewed material with dielectric changes, particularly those due to water content, and with measurement frequency. Knowledge of this behaviour is essential for estimating, at one extreme, relatively small but physiologically important temperature gradients within the human body, and at the other extreme, the large and rapidly varying temperature patterns induced during industrial processes. By measuring the microwave temperature at different microwave frequencies, it is possible to retrieve information on the temperature at varying depths within the material. To aid in the interpretation of these measurements, the BEST program was used to ascertain the form of the weighting function at two frequencies, 1.35 GHz and 3.2 GHz, for a specific dual - frequency antenna in a range of phantom materials. The phantom materials were composed of a mixture of water, protein and salts, with the intention being to simulate common biological materials. To consider foodstuffs a mashed potato phantom was used. Chapter 5 includes the design of this dual frequency antenna and its application to measuring the radiometric temperature of non-isothermal mashed potato mixtures. The specific manipulation of the potato mixture (through heating and cooling) to produce known temperature profiles (quasi-linear and quasi-quadratic) is also considered in this chapter. Further validation of the BEST weighting function determination is possible by comparison with these experimental temperature measurements. Chapter 6 initially covers the modelling of the dielectric properties of the mashed potato and protein / saline mixtures. In particular, a model of the variation of the dielectric constant and loss factor of the mashed potato material, covering a wide range of temperatures at 1.35GHz and 3.2GHz, is presented and shown to agree with published literature. The effects on the computed weighting function of variation of several key factors, including measurement frequency and material temperature, are then considered for both phantom types. Further, limitations in the computational modelling in terms of finite bounds and the modelling of layers are investigated. Finally, techniques for obtaining the physical temperature distribution from multi - frequency microwave readings are considered in chapter 7 and their applicability at two frequencies is discussed. By making use of the data collected from the dual - frequency antenna and simulated microwave temperatures, the various methods of temperature profile retrieval are compared.