A Novel Tumour Characterization in Microwave Imaging Using Pattern-Based Weighted DMAS

The usage of microwave imaging for biomedical (BMWI) applications is still challenging due to the imprecise reconstruction of the relative permittivity ($\varepsilon_{\text{r}}$) of tissues and the ill-posed inverse scattering problem. Here, a novel two-step approach is proposed for tumour characterization in the BMWI framework. The focus is distinguishing normal and tumourous tissues using a non-iterative approach. An intensity map of the target region is generated to detect the tumour. Then the area around the tumour is further processed to obtain the refined image in terms of the effective $\varepsilon_{\text{r}}$. As the tumour is embedded in tissues of higher dielectric constant, the image thus formed will usually be larger in size than the actual tumour. Thus, to estimate the accurate size, we propose a novel antenna pattern-based weighted delay, multiply, and sum algorithm along with principal component analysis as a synthetic focusing technique. Subsequently, the effective dielectric profile of the embedding medium is approximately estimated, which is close to their actual values.