Experimental Evaluation of a Microwave Tomography System for Breast Cancer Detection
Microwave tomography is a potential candidate for future breast-cancer screening or diagnosis. Contrary to x-rays, microwaves are non-ionizing and therefore not a health risk by their own. The examination procedure would also be more comfortable for the patient compared to conventional mammography since no compression of the breast is needed.
The examination is performed by irradiating the breast with microwaves from multiple directions. The collected data is then processed by an iterative algorithm that reconstructs the permittivity and conductivity distribution in the interrogated region. Ideally, tumors could be identified in these reconstructed images due to their different properties compared to normal tissue.
In this thesis, a prototype system for microwave tomographic imaging is experimentally evaluated. The system consists of 16 monopole antennas and utilize a mixture of water and glycerin as coupling liquid. As a tool for the assessment, two phantoms have been studied. One is a simplistic phantom consisting of a cylinder in which smaller cylindrical inclusions can be inserted. The other is a 3D printed structure made to resemble a human breast geometrically. This particular phantom consists of two shells, representing the different tissues of the breast. The system is found to produce well reconstructed images of both the interrogated phantoms. However, the interior geometry of the 3D printed phantom was more challenging.
Furthermore, two different reconstruction algorithms are tested. The first is a Gauss-Newton based FEM algorithm and the second is a gradient-descent based FDTD method. Both of the studied algorithms proved to yield good reconstructions.