Unsupervised deep learning with convolutional neural networks for static parallel transmit design: A retrospective study.

Proposed unsupervised deep learning with CNNs performs better than unregularized MLS in static pTx for speed and robustness.