A novel strategy to guarantee convergence by avoiding divergence in estimation process for evaluating head tissue conductivities

Abstract

One of auxiliary processes to obtain the optimal solution of an inverse problem is to use guaranteeing convergence strategy. Line-search strategy is the oldest and the most widely used one. However, line-search strategy is not suitable in some situations, in particular, in nonlinear situations where many local minima are present. Seeking an optimal solution for head tissue conductivities estimation with electrical impedance tomography (EIT) technique is a kind of such situation. Noise and modeling error also are crucial factors to determine the accuracy and the reliability of the estimation, and, many times, the line-search strategy cannot deal with. In this work, a novel guaranteeing convergence strategy is proposed by restricting the change of estimates during the iterative process of the estimation. Respecting the simulation result, the novel strategy employment shows the improvement of reliability and local minima avoidance even working in a high-degree nonlinear situation.