Manyetik rezonans elektriksel empedans tomografisi için optimum akım uygulama stratejisi.


Tezin Türü: Doktora

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye

Tezin Onay Tarihi: 2008

Tezin Dili: İngilizce

Öğrenci: Haluk Altunel

Danışman: BEHÇET MURAT EYÜBOĞLU

Özet:

In this thesis, optimum current injection strategy for Magnetic Resonance Electrical Impedance Tomography (MREIT) is studied. Distinguishability measure based on magnetic flux density is defined for MREIT. Limit of distinguishability is analytically derived for an infinitely long cylinder with concentric and eccentric inhomogeneities. When distinguishability limits of MREIT and Electrical Impedance Tomography (EIT) are compared, it is found that MREIT is capable of detecting smaller perturbations than EIT. When conductivities of inhomogeneity and background object are equal to 0.8S and 1S respectively, MREIT provides improvement of %74 in detection capacity. Optimum current injection pattern is found based on the distinguishability definition. For 2-D cylindrical body with concentric and eccentric inhomogeneities, opposite drive provides best result. As for the 3-D case, a sphere with azimuthal symmetry is considered. Distinguishability limit expression is obtained and optimum current injection pattern is again opposite drive. Based these results, optimum current injection principles are provided and Regional Image Reconstruction (RIR) using optimum currents is proposed. It states that conductivity distribution should be reconstructed for a region rather than for the whole body. Applying current injection principles and RIR provides reasonable improvement in image quality when there is noise in the measurement data. For the square geometry, when SNR is 13dB, RIR provides decrement of nearly %50 in conductivity error rate of small inhomogeneity. Pulse sequence optimization is done for Gradient Echo (GE) and it is compared with Spin Echo (SE) in terms of their capabilities for MREIT.