Antenna designs and imaging algorithm for radar-based microwave breast cancer detection
[ACCESS RESTRICTED TO THE UNIVERSITY OF MISSOURI AT AUTHOR'S REQUEST.] In this work, the design of ultra-wideband antennas and an algorithm for artifact removal and image reconstruction beamforming for microwave breast cancer detection were presented. A tapered slot antenna and a quad-ridge horn antenna are designed, simulated and tested, and a system for multistatic radar-based microwave tumor detection for both artifact removal and image reconstruction was proposed. The system can be fabricated as a portable electronic chip. This work incorporates antenna features required for radar-based microwave breast cancer detection through the design of an ultra-wideband tapered slot antenna embedded in a dielectric material and testing the antenna in the time domain, frequency domain and in imaging reconstruction. The analysis shows impedance matching over ultra-wide frequencies while maintaining a high gain with low distortion. Two imaging approaches were also proposed. Results validate the proposed design and the imaging approaches. A dual polarized quad-ridge horn antenna was also proposed, in which each flared ridge was fitted with a semi-elliptical metallic part to increase bandwidth and polarization isolation. Vertical and horizontal oriented field probes were used to test isolation between the polarization components. Results show an almost 40 dB isolation between the linear polarization signals with high gain and port to port isolation over the entire bandwidth. Following the excitation signal, the quad-ridged horn antennas show unwanted signals that persist for a long time. This interference needs to be removed for successful detection of the cancerous growth. Analysis shows that the sources of oscillating were from the antenna cavity resonance and the inherent antenna LC resonator. Solutions to remove or minimize these signals without affecting the antenna parameters were implemented. Modification of the cavity successfully suppressed the cavity oscillation and altering the antenna waveguide reduced inductance and thus mitigated LC oscillation. The time and frequency domain signals demonstrated the effectiveness of the proposed solutions. Furthermore, the proposed techniques enhanced image quality through clutter reduction. The proposed radar-based microwave breast cancer detection system removed undesirable artifacts from multistatic signals reflected from the breast. A heterogeneous breast phantom, with two tumors, were simulated. After the multistatic scan, the collected responses were divided into groups of highly correlated signals. A window function was derived from the signals to remove parts which had no target information. The signals in each group were processed using Wiener filter, and the processed signals of all the groups are then merged into one artifact-free data matrix. The signals from different scans were combined in the beamforming step. An improved beamforming imaging algorithm was also proposed. Results demonstrated the merits of the proposed method in removing the artifact in multistatic radar situation. Finally, the images demonstrated the effectiveness of the proposed algorithm in minimizing clutter and in determining the actual size of a breast tumor with low computational complexity.
Access to files is limited to the University of Missouri--Columbia.