Deep learning-based local SAR prediction using B-1 maps and structural MRI of the head for parallel transmission at 7 T

Purpose: To predict subject-specific local specific absorption rate (SAR) distributions of the human head for parallel transmission (pTx) systems at 7 T. Theory andmethods: Electromagnetic energy deposition in tissues is nonuniform at 7 T, and interference patterns due to individual channels of pTx systems may result in increased local SAR values, which can only be estimated with very high safety margins. We proposed, designed, and demonstrated a multichannel 3D convolutional neural network (CNN) architecture to predict local SAR maps as well as peak-spatial SAR (ps-SAR) levels. We hypothesized that utilizing a three- channel 3D CNN, in which each channel is fed by a B-1(+) map, a phase-reversed B-1(+) map, and anMR image, would improve prediction accuracies and decrease uncertainties in the predictions. We generated 10 new head-neck body models, along with 389 3D pTxMRI data having different RF shim settings, with their B-1 and local SAR maps to support efforts in this field. Results: The proposed three-channel 3D CNN predicted ps-SAR(10g) levels with an average overestimation error of 20%, which was better than the virtual observation points-based estimation error (i.e., 152% average overestimation). The proposed method decreased prediction uncertainties over 20% (i.e., 22.5%-17.7%) compared to other methods. A safety factor of 1.20 would be enough to avoid underestimations for the dataset generated in this work. Conclusion: Multichannel 3D CNN networks can be promising in predicting local SAR values and perform predictions within a second, making them clinically useful as an alternative to virtual observation points-based methods.

Automated summary

The purpose of this study is to predict subject-specific local specific absorption rate (SAR) distributions of the human head for parallel transmission (pTx) systems at 7 T. A multichannel 3D convolutional neural network (CNN) architecture was proposed, designed, and demonstrated to predict local SAR maps as well as peak-spatial SAR (ps-SAR) levels. The results showed that the proposed method had smaller prediction errors and could make predictions within a second, making it clinically useful.