Patient-Specific RF-induced Heating Prediction for Deep Brain Stimulator System under Parallel Transmit at 3T with Machine Learning Method

The magnetic resonance imaging (MRI) radio frequency (RF)-induced heating of the deep brain stimulators (DBSs) is dependent on the patient’s characteristics and device implantation. The rule is further varied under parallel transmission. The traditional RF-induced heating evaluation method is comprehensive but time-consuming. Hence, an artificial neural network (ANN) model is developed in this study to predict the RF-induced heating for DBS systems under the parallel transmit condition at 3T. The hyperparameters of the ANN model, including the learning rate and the number of hidden layers and neurons, are optimized to handle the dataset of 66,290,400 samples. The original data is generated with the traditional transfer function method. The performance of the ANN model is validated with four different transfer functions. Using the optimized ANN regression model, the RF heating of the DBS system can be quickly and accurately predicted. The performance of the ANN regression model has a mean absolute error (MAE) of less than 0.84 °C and a coefficient of determination (R²) value of 0.996.