Noise-Robust Microwave Breast Imaging Applied to Multi-Frequency Contrast Source Inversion

摘要

Ultra-wideband (UWB) microwave quantitative imaging offers less-painful examination, where a significant dielectric contrast between the malignant tumor and normal tissue is exploited. However, there are some difficulties in providing an accurate dielectric profile of breast media. This is because additive noises severely contaminate scattered signals. In this study, we apply a post-processing multi-frequency integration scheme to contrast source inversion (CSI) data to suppress image fluctuations. These are caused by the additive noise expected in a UWB system, which is also applicable to other inversion schemes. To deal with the dispersive dielectric properties in the CSI scheme, we introduce a first-derivative model of the Debye dispersion model to compensate for the dispersive effect. The FDTD numerical validations, using realistic breast phantoms with dispersive properties, show that a multi-frequency integration scheme considerably upgrades noise-robustness in complex permittivity reconstruction tissues.