Brain Stroke Localization Using Four Layered Head Phantom

Abstract

Nowadays, brain stroke is commonly occurring among many people regardless of age, gender, race or region. In order to diagnose the exact location and size of a blood clot (stroke), brain imaging systems are required. However, conventional imaging systems including MRI, CT, and X-ray for stroke localization are expensive and located at hospitals or physical examination centers. As a result, the demand for new brain imaging systems, which are fast responding, economical and portable, has increased especially in rural areas or war zones. Several research groups have developed new brain imaging systems for stroke localization based on microwave systems since the microwave systems are relatively more economical and safer than ionizing devices (CT, and X-ray) and MRI. To develop the microwave imaging system, a realistic phantom for validation test, an array of miniaturized antennas, and proper imaging algorithms are needed. In this dissertation, a miniaturized antenna and brain imaging algorithm for microwave brain stroke localization system and a realistic head phantom for validity evaluation test bed are proposed. For evaluating the validity of a brain imaging system, brain phantoms are used instead of a real human head due to safety issues caused by electromagnetic wave exposure. To perform the realistic evaluation test simulating the actual situation, the evaluation test bed requires a brain phantom whose anatomical structure and electrical properties are identical to those of a human head. However, previous research has used a phantom filled with a single material or a multi-layered phantom with a ready-made brain mold (i.e the shape of a skull fixed for all patients). In this dissertation, a four layered brain phantom for evaluating the validity of a brain stroke system is proposed. The mold for the proposed phantom is printed by a 3D printer and the interior of the phantom consists of 4 different brain tissue materials and blood material simulating blood clots caused by a stroke. Each of the brain tissue materials has the conductivity and permittivity similar to those of ICNIRP standards for a frequency band from 0.5 GHz to 2 GHz. In order to enhance the performance of a brain stroke localization device, the antenna should have wideband (0.5Ghz to 2Ghz) characteristics since a high dielectric constant and conductivity of a brain tissue disturb the propagation of microwave pulse signals. In addition, the size of an antenna should be small to form a compact array system surrounding the human head. To meet the requirements, an elliptical monopole antenna having near omnidirectional and broadband radiation characteristics is chosen as a basic element. To reduce the size of the antenna, slots are embedded on its radiating patch. The slots do not only reduce the size but also generate unwanted bandstop characteristics at frequencies above 1.7 GHz. To resolve the aforementioned problem, high pass filters are designed and located at the end of each slot. As a result, the antenna properly operates without having bandstop characteristics at higher frequencies. An imaging algorithm for brain stroke localization is needed to figure out the location of the object of interest (blood clot) surrounding high dielectric materials (CSF or white matter) with a short processing time. One of the imaging algorithms used for brain stroke localization is microwave tomography (MT) which is contrasting images of dielectric constants forming received signals at the antennas since electrical properties of blood are significantly different from those of other brain tissues. One research group developed MT to eliminate background signal scattered by the skin of the head by subtracting the two received signals from antennas located nearby. In addition, the MT calculates minimum electrical length to localize the scatterer from received signals. However, the MT assumes the geometry of a skull as an ellipsoid, combining two halves of ellipses with different sizes, which is different form practical cases. As a result, the previous method result has marginal error since different propagation speeds are applied inside and outside of the head. In this dissertation, geometry of the head is detected from received signals and used for calculating the minimum electrical length. Firstly, distance from the antenna to the head is detected from received signals before eliminating the background signal. After that, boundary points are calculated from the distances between the antennas and head. In addition, the shape of skull is estimated by interpolating the boundary positions. Six stroke scenarios (radius : 0.5, 1, and 2 cm, and depth : 4 and 8 cm) are considered to validate the proposed algorithm and are simulated with a 3D EM simulator. And the microwave signal is processed with two different methods. In order to verify the validity of the proposed localization method, the same stroke scenarios are analyzed using the proposed method and the conventional method. One is the proposed algorithm and the other method is the conventional method. In addition, a test bed for the proposed brain imaging system is constructed to perform the measurement. The test bed consists of a 4-layered phantom, transmitting and receiving modules including the proposed elliptical antenna, and a jig. Measurements are also performed for the aforementioned scenarios using two different methods. The results from simulations and measurements are compared. The comparison between the proposed and conventional stroke localization methods reveals that the proposed localization algorithm more precisely indicates the stroke location.