Rapid estimation of 2D relative B1+ -maps from localizers in the human heart at 7T using deep learning
Felix Krueger, Christoph Stefan Aigner, Kerstin Hammernik, Sebastian Dietrich, Max Lutz, Jeanette Schulz-Menger, Tobias Schaeffter, Sebastian Schmitter
Purpose
Subject-tailored pTx pulses for ultra-high fields body applications are typically calculated based on subject-specific B1±maps of all transmit channels, which requires lengthy adjustment times. This study investigates the feasibility of utilizing deep learning to estimate complex, channel-wise, relative 2D B1±maps from a single gradient echo localizer to overcome long calibration times.
Methods
126 channel-wise, complex, relative 2D B1±maps of the human heart from 44 subjects were acquired at 7T using a Cartesian, cardiac gradient-echo sequence obtained under breath-hold to create a library for network training and cross-validation. The deep learning predicted maps were qualitatively compared to the ground truth. Phase-only B1±shimming was subsequently performed on the estimated B1±maps for a region of interest covering the heart. The proposed network was applied at 7T to three unseen test subjects.
Results
The deep learning-based B1±maps, derived in approximately 0.2 seconds, match the ground truth for the magnitude and phase. The static, phase-only pulse design performs best when maximizing the mean transmission efficiency. In vivo application of the proposed network to unseen subjects demonstrates the feasibility of this approach: The network yields predicted B1±maps comparable to the acquired ground truth and anatomical scans reflect the resulting B1±pattern using the deep learning-based maps.
Conclusion
The feasibility of estimating 2D relative B1±maps from initial localizer scans of the human heart at 7T using deep learning is successfully demonstrated. Since the technique requires only sub-seconds to derive channel-wise B1±maps, it offers high potential for advancing clinical body imaging at ultra-high fields.