Hybrid Method for Image Reconstruction in Magnetic Induction Tomography: Combination of the Neural Network with an Iterative Analytical Algorithm

Abstract Magnetic induction tomography (MIT) is an imaging technique that uses eddy current effects to map the conductivity of low conductive objects. It has a wide range of applications and offers numerous possibilities in various fields. One task of this imaging method involves resolving the ill-posed and nonlinear inverse problem. However, existing methods have weaknesses. A hybrid approach that integrates theoretical algorithms with machine-learning methods is proposed as an innovative solution to the challenges of MIT reconstruction.