Prediction of Active Implantable Medical Device Electromagnetic Models Using a Neural Network

In this paper we use a neural network (NN) to develop an Active Implantable Medical Device (AIMD) electromagnetic model for Magnetic Resonance (MR) exposure. The tangential electric (E)- fields along the AIMD leads are chosen as input for the NN and the RF-induced header voltage is chosen as the NN output. Fourteen hundred data points are used for network training, validation, and testing. Results show good convergences during network training with a worst mean absolute error (MAE) of 30% for the developed lead model.