| Electrical impedance tomography is a technology for producing images of internal body structures based upon electrical measurements made from electrodes on the body surface. Mathematically, we must solve an inverse problem. That is, apply currents on the surface of a body, measure the resulting voltages, and then determine what the conductivity distribution must be based upon the applied currents and measured voltages. Typically a single plane of electrodes is used in an attempt to reconstruct a cross-section of the body, and the majority of previous algorithms ignore the three-dimensional (3D) characteristics of the current flow in the body. Actually, only a relatively small percentage of the current remains in the electrode plane, with the out-of-plane currents creating distortions in the resulting images.; This thesis discusses the design and implementation of a reconstruction algorithm, ToDLeR, for solving the linearized 3D inverse problem in impedance imaging. The algorithm models the body as a homogeneous cylinder and accounts for the 3D current flow in the body by analytically solving for the current flow from one or more layers of electrodes on the surface of the cylinder. The real-time implementation of the algorithm in the ACT3 Real-Time Imaging System, which displays approximately 21 images per second, is described. Data was collected from a 3D test phantom using one, two, and four layers of electrodes. By using multiple planes of electrodes, improved accuracy in any particular electrode plane was obtained, with decreased sensitivity to out-of-plane objects.; The real-time system was also used to collect data from the thorax of a human volunteer using a 4-layer electrode configuration with 8 electrodes per layer. Reconstructed images are presented from sequences of both cardiac and respiratory events. The cardiac sequence shows the phasic nature of blood flow between the heart and the pulmonary circulation; as the conductive blood is pumped from the heart, a heart region becomes less conductive, while simultaneously the regions of the lungs become more conductive. Furthermore, different anatomical structures are seen in different levels of the reconstructions, as would be expected given the 3D nature of the human thorax. We conclude that the 3D algorithm, used with multiple planes of electrodes, minimizes the distortions from out-of-plane structures in the body. |
