We propose a deep learning method for abdominal electrical impedance tomography (EIT) to estimate abdominal obesity. EIT for evaluating abdominal obesity is a challenging problem that is an ill-posed absolute imaging problem. The proposed method allows to find an useful solution within a restricted admissible set, accounting for prior information on abdominal anatomy. It found that a specially designed training data used in the deep learning process reduces the ill-posedness in the absolute EIT problem.
This presentation is part of Minisymposium “MS54 - Hybrid Approaches that Combine Deterministic and Statistical Regularization for Applied Inverse Problems (4 parts)”
organized by: Cristiana Sebu (University of Malta) , Taufiquar Khan (Clemson University) .