Microwave Tomography Using Random Multiple-Input–Multiple-Output Contrast Source Inversion

Microwave tomography using the contrast source inversion (CSI) method has been intensively researched especially in medical applications, where a coupling medium (CM) is usually used to improve imaging. Previous works noted that better results can be achieved when the CM is lossy. Nevertheless, lossy CM weakens the target response, especially when complicated scenario is imaged such as the head. In this letter, CSI is analyzed to indicate that the dielectric properties of CM are related to the condition number of a part of the objective function in each of CSI iteration. It is found that the value of this condition number is proportional to the ratio of maximum and minimum electric field intensity in the computational domain. Therefore, to reduce that ratio, a random multiple-input-multiple-output (RMIMO) scheme, which is implemented by applying a transformation matrix on the measured S-parameter matrix, is proposed to enable the use of loss-less CM and improves the imaging results. The proposed analysis and RMIMO-CSI method are validated using numerical head models.