A computational study of reconstruction algorithms for ultra-wideband electromagnetic pulse imaging in tissue

Microwaves can be transmitted through, absorbed by, and reflected at biological tissue boundaries in varying degrees. These factors enable microwaves to be used for diagnosis in medicine, especially for the imaging of the biological structures depending on the tissue properties and frequency. Breast tissue shows relatively penetrable characteristics up to several gigahertz frequency range because of its fatty composition. Also, its small volume makes it amenable for microwave imaging applications. Although its heterogeneous nature complicates imaging, the high electrical property value (permittivity and conductivity) contrast of normal and malignant human breast tissue suggests active microwave imaging methods are promising for the detection of tumors in breast tissue.; This dissertation has developed a methodology to detect and identify a tumor in breast tissue in 3-D computer simulations with the FDTD method. We have developed an algorithm to detect the tumor location in the breast tissue in terms of its relatively high microwave scattering property within the surround tissue for a particular breast and imaging-system model. A data bank has been created to model the returns of different type and shape of tumors to be used in a matched-filtering test based on the tumor location information. The performance of the matched-filtering in terms of the tumor size and antenna-array size has been demonstrated. Finally, a new approach has been proposed to optimally reduce the clutter level in the breast return signals since this is the biggest obstacle in the performance of the matched-filtering. We have shown that this approach reduces the burden on the hardware for successful matched-filtering conditions.