Simulation and Analysis of Two Point-like Scatterers in a Portable Microwave Breast Cancer Detection System

Breast cancer screening plays an essential role in the early detection of breast cancer, but access to screening is limited for women in remote locations and low-income countries. Microwave imaging systems address this disparity, offering a compact, cost-effective approach that uses safer non-ionizing radiation to detect breast cancer. To optimize and calibrate a novel portable microwave breast cancer detection device, this study focuses on characterizing changes in the electromagnetic (EM) fields associated with Aluminum (Al) rods serving as point-like scatterers.A replica of an experimental portable microwave breast cancer detection system was designed in CST Microwave Studio software. This system used a rotating horn antenna and a hemispheric array of point-like sensors surrounding the breast to detect the EM fields scattered by the breast. The signal detected by these sensors is compared to the results from a normal breast to detect the presence of a malignant lesion. The simulation and analysis followed the methodology presented in earlier single rod work, where the ratios (R) of the magnitude of the EM fields with and without the rod(s) were calculated. The analysis was conducted over a bandwidth of 2 to 8 GHz for two Al rods at different locations and spacings within the imaging chamber. The magnitudes of the E-fields with the rods were compared to the magnitudes of the E-fields for a single Al rod placed at similar positions.The two-rod response can be estimated from a weighted non-linear combination of the single-rod responses. This supports the hypothesis that the results for multiple point-like scatterers can be obtained from the superposition of many single rod responses. While further analysis is needed to determine the results of more complex breast models (with multiple scattering locations), this work enables us to optimize the antenna locations to maximize system sensitivity and specificity, provides the basis for a computationally efficient method of generating data for training a machine learning network, and provides a reliable technique to calibrate our physical system.