Convex optimization approach for stroke detection in microwave head imaging

Convex optimization provides a method of minimization of a convex objective function subject to a convex domain imposed upon it by the problem. For microwave imaging in medical applications, such as head imaging, this technique is seldom investigated. In this paper, a microwave-based head imaging method based on convex optimization is presented. Convex optimization is used to successfully estimate the distribution of relative permittivity of the imaged objects at different directions and thus to improve the quality of the obtained microwave image. The obtained results using 32 antennas surrounding a realistic head model compare favorably with the images from using the traditional microwave head imaging algorithm, which assumes a certain fixed average permittivity for the whole imaged head. The results show that the target representing a bleeding inside the head is properly recovered using the proposed optimization despite using wide range of initial average permittivity values. However, the quality of images produced using the traditional approach depends strongly on the assumed average permittivity.