Authors : G.K. Aleksanyan, P.A. Denisov, N.I. Gorbatenko, I.D. Shcherbakov and I.S.D. Al Balushi
Abstract: A complex nonlinear dependence of the electrical parameters on the injecting current frequency that differs for different tissues, leads to the need of use several mathematical models: three, five and six component. Since, it's impossible to reliably determine exact location of the tissue types boundary in the body before measurement, there is a problem of suitable mathematical models selection in the multi-frequency electrical impedance tomography. In order to overcome mentioned difficulties method of biological tissues identification based on natural-model approach, applicable for two-dimensional and three-dimensional multi-frequency electrical impedance tomography is proposed. In this case injecting current of different frequencies can be supplied either simultaneously or at different times. It is assumed decomposition mathematical model of a biological object into two levels of simulation: distribution of conductivity, dielectric constant and currents in the area occupied by the object and model of object electrical parameters depending on injecting current frequency. The proposed method allows to select the most adequate model of dependence biological tissues local electrical parameters on injecting current frequency, it would be useful for the automatic identification of tissue located in considerated area. Model selection algorithm is based on the principle of building hypotheses and selection of them as a result of subsequent measurements that most accurately predict the electrical parameters dependence on frequency. The technique involves a measurement for the pre-selected injecting current frequencies, followed by automatic detection of an additional set of frequencies and additional measures for them. Thus, the scheme of cyclic multi-frequency electrical impedance tomography is proposed. At each iteration of the loop are executed sequentially: the injection of current and potential value measurements, reconstruction of the distribution of the electrical parameters, the analysis of parameters dependence on the frequency and the selection of the most adequate model. Reconstruction of the electrical parameters distribution can be performed using known numerical techniques for solving inverse problems of identification. An analysis of the applicability of the stationary and symbolic models currents spreading for the implementation of the proposed approach shows that even in the region of stationary model applicability, it's desirable to use a symbolic model due to the fact that it allows to get information about dielectric constant distribution.
G.K. Aleksanyan, P.A. Denisov, N.I. Gorbatenko, I.D. Shcherbakov and I.S.D. Al Balushi, 2018. Principles and Methods of Biological Objects Internal Structures Identification in Multifrequency Electrical ImpedanceTomography Based on Natural-Model Approach. Journal of Engineering and Applied Sciences, 13: 10028-10036.