Abstract:

Stroke is one of the main causes of death in western society. Regarding the two types, hemmorrhagic or ischemic stroke, fundamentally different treatment is necessary. In diagnosis of stroke, several imaging technologies like CT or MRT are already established but very expensive und not transportable. Microwave tomography is a promising technology to create a solution in order to improve diagnosis of hemorrhagic stroke. The developement of a measurement setup utilized for stroke detection is done using FDTD- Simulation to test the design and antennas. In this work, the approach was performed using a rectangular antenna-array with Vivaldi antennas for measurement. Therefor, a simple but realistic head-phantom with permittivity values comparable to human tissue was used. The simulation and measurement results are calculated by a software framework. It includes forward calculation using the MEEP-Framework and Gauss- Newton optimization for the update of values. The ill-posed inverse Problem is regularized by Tikhonov algorithm.

Abstract:

Stroke and Traumatic Brain Injury is one of the main causes of death in modern society. Regarding hemorrhagic stroke and other intracranial bleeding, fast and mobile imaging technology is in focus of development to support reliable information on the patients disease. Microwave tomography is a promising technology for future use in intracranial diagnostics. Beside the abdication of X-rays, this development should be mobile and pro- vide real-time measurement supporting the right decision for the correct care.The measurement of complex permittivity values is the recommended information for final image results. Beside the measurement setup, development of image reconstruction algorithms and optimization of nonlinear solvers are in focus of current research. In this project, Gauss-Newton-Optimization is used for approximation. MEEP is used for the forward calculation of the propagating field. Regarding an existing framework, PETSc is implemented for the most important software features like, preconditioning and Gauss- Newton implementation with line search algorithm as well as regularization.Hardware optimization is done by a change of the measurement setup, like antenna po- sitions, number of antennas, amplifiers and the first-guess model. An outlook for further improvements is given.