Improved Resolution and Reduced Clutter in Ultra-Wideband Microwave Imaging Using Cross-Correlated Back Projection: Experimental and Numerical Results

摘要

Microwave breast cancer detection is based on the dielectriccontrast between healthy and malignant tissue. This radar-basedimaging method involves illumination of the breast with anultra-wideband pulse. Detection of tumors within the breast isachieved by some selected focusing technique. Image formationalgorithms are tailored to enhance tumor responses and reduceearly-time and late-time clutter associated with skin reflectionsand heterogeneity of breast tissue. In this contribution, weevaluate the performance of the so-called cross-correlated backprojection imaging scheme by using a scanning system in phantomexperiments. Supplementary numerical modeling based on commercialsoftware is also presented. The phantom is synthetically scannedwith a broadband elliptical antenna in a mono-staticconfiguration. The respective signals are pre-processed by adata-adaptive RLS algorithm in order to remove artifacts caused byantenna reverberations and signal clutter. Successful detection ofa 7 mm diameter cylindrical tumor immersed in a low permittivitymedium was achieved in all cases. Selecting the widely useddelay-and-sum (DAS) beamforming algorithm as a benchmark, we showthat correlation based imaging methods improve thesignal-to-clutter ratio by at least 10 dB and improves spatialresolution through a reduction of the imaged peak full-width halfmaximum (FWHM) of about 40–50%.