Adapting Seismic Processing Techniques for Data Preconditioning in Radar Imaging of Highly Dissipative and Dispersive Media

摘要

The concept of using microwave frequency electromagnetic waves for biomedical imaging applications has interested researchers for decades. Promising results have been reported for several approaches to microwave breast imaging, including radar-based imaging applied to realistic numerical breast phantoms and patient studies. However, important problems have also been identified, specifically, low image resolution and sensitivity due to multiple-scattering effects and frequency-dependent attenuation in the presence of highly dissipative and dispersive breast tissues.;Microwave imaging and seismic imaging deal with analogous problems. In seismic imaging, tremendous efforts have been invested in developing data analysis and preconditioning techniques to render the accurate graphical representation of specific portions of the earth's subsurface geological structure. The overall objective of this thesis is to produce more accurate microwave breast images from ultra-wideband radar signals by adapting advanced seismic imaging techniques. First, we develop a method based on first-breaks to detect the pulse arrival time in the presence of severe waveform distortion. Second, we adapt Gabor nonstationary deconvolution to accurately estimate the subsurface reflectivity in the presence of severe attenuation and dispersion due to EM wave propagation in highly lossy dispersive biological tissues at microwave frequencies. Third, we develop a dual deconvolution processing flow (DDPF) to account for the interfering responses present in a radar reflection measurement system.;The proposed methods are applied to simulated and measured data. The results indicate that the first-break time is able to provide consistent and reliable reference for travel time estimation in the presence of severe waveform distortion and Gabor deconvolution is able to effectively compensate for wave attenuation in highly lossy and dispersive media. The preliminary imaging test demonstrated a significant improvement in the image sensitivity with Gabor deconvolution preconditioned data. Application to the simulations of realistic breast phantoms and experimental patient scans shows that the DDPF method is able to detect the scatterers in the presence of heterogeneous, lossy, and dispersive tissues. Overall, this study demonstrates successful modification of seismic data preconditioning techniques to biomedical radar data, resulting in images with improved accuracy.