A Novel B1+ Mapping Method for Deuterium MRI
Abstract
Purpose
A novel high-resolution and bias-free deuterium MRI B1+ mapping method was developed for validating the simulation accuracy of a 7T body RF coil. The extremely low SNR of X-Nuclei causes significant problems for conventional B1+ mapping techniques, resulting in long acquisition times, low resolution, and biased estimates. Here, we describe a novel approach which uses parametric modelling of multi-echo gradient-echo (ME-GRE) data, acquired as a function of coil transmit voltage, to extract estimates of the B1+ value at each pixel. B1+ values are compared to a conventional mapping approach using chemical shift imaging (CSI) and processed with AMARES.
Methods
ME-GRE experiments were performed in an abdomen-sized phantom containing 280mL D2O, at 7T (Terra.X, Siemens) using an 8-Tx/Rx (2H) body array coil, with: voxel=9x9x50mm^3, TR/TE=2500/5-80ms(12-steps), NSA=10, voxel=9.2x9.2x25mm3, Transmit_V=10.34-300V(30-steps), acquisition-time=203mins. CSI experiments were performed with: TR/TE=3000/2.3ms, NSA=18, BW=1kHz, voxel=12.5x12.5x25mm3, Transmit_V=18.5-370V(20-steps), acquisition-time=516mins. Data was processed using custom scripts in MATLAB (MathWorks). Parametric processing used the Steiglitz-McBride algorithm with sub-band decomposition to remove noise and determine magnetization at each transmit voltage. Voxel-wise fitting to a sine curve yielded B1+ values. Simulations were performed using sim4life (ZMT, Switzerland).
Results
Figure 1 compares B1+ maps from CSI and ME-GRE, demonstrating the significantly improved SNR and spatial resolution achieved with reduced acquisition time. The spatial distribution of the ME-GRE B1+ map closely matched that of the simulated map. Figure 2 shows that the CSI-AMARES method had a weak correlation with simulated data when compared voxel-wise, and furthermore introduced a bias error. The correlation with the ME-GRE method was significantly improved.
Conclusion
The novel ME-GRE B1+ mapping sequence achieves high spatial resolution and accurate B1+ estimates in reasonable acquisition times, and facilitates accurate validations of simulation models for newly-developed RF coils. The technique can be easily adapted to other X-Nuclei.