Imaging methods at high fields can suffer from receive non-uniformities from the body coil, particularly when the body coil is used as a reference for intensity correction. The B₁ mapping method DREAM is widely used for rapid mapping of the transmit B₁⁺ field.¹ Recently, it was shown that also the receive sensitivity (B₁^-) can potentially be retrieved from DREAM data.² Using this approach, we evaluate the use of DREAM for correcting the receive non-uniformities of the body coil in whole body MRI at 3T.

A whole-body imaging protocol was performed at different landmarks in a 3T dual-transmit MR system (Ingenia, Philips Healthcare, Best, the Netherlands). The two channels of the body coil were driven in RF-shimmed mode, using the vendor supplied RF-shimming routine. The DREAM method was used for multislice B₁ mapping; in-plane resolution = 7 × 7 mm², slice thickness = 7 mm, (nominal) STEAM/imaging tip angle = 10°/50°, TR/TE_ste/TE_fid=3.8/1.05/2.3 ms. Receive sensitivity (B₁^-) maps were obtained via the DREAM data as derived in Eqs. [1-3]. The B₁^- profiles were then smoothened by means of a regularized polynomial fitting procedure to remove density weighting.

$$I_{\rm FID}=M_0 B_1^- \sin\beta \cos^2\alpha \left[1\right]$$

$$I_{\rm STE}=M_0 B_1^- \sin\beta \sin^2\alpha \left[2\right]$$

$$M_0 B_1^- = \frac{I_{\rm FID}+2I_{\rm STE}}{\sin\beta} \left[3\right]$$

Two-point mDIXON images were acquired using the body coil for signal reception via a dual-echo 3D gradient echo sequence; voxel size = 1.5×1.5×10 mm³, TR/TE1/TE2 = 3.6/1.2/2.0 ms, tip angle = 5°. T2-weighted Fast Spin Echo (FSE) images were acquired using a torso receive array; voxel size = 1.3×1.3 mm²; slice thickness = 5 mm; slice gap = 5 mm; TR/TE = 1000/80 ms; excitation/refocusing tip angle = 90°/120°; 12 refocusing pulses; sensitivity encoding (SENSE) acceleration = 2.

Fig. 1 shows reconstructed B₁⁺ and B₁^- profiles from the DREAM sequence based on the single-slice calibration data combined to mimic quadrature driving conditions. The results show that there is a mirror-symmetry relation between the transmit and receive profiles, which is in agreement with basic electromagnetic theory.³

In Fig. 2 RF-shimmed multislice DREAM data is used to correct the in-phase images of the reconstructed DIXON data. Improved image uniformity is especially visible in the torso and legs. It is also worth noting that the image uniformity is improved at the cardiac station, despite the flow-sensitivity of the DREAM method.

Fig. 3 shows FSE images in the liver and lower extremity stations, before and after correction. Although less visible due to the strong T2 weighting, image uniformity is improved in both examples. Window levels have been adjusted to the level of the liver and muscle to better visualize the improved uniformity.

This study shows that DREAM data can be used to perform uniformity correction in body applications at 3T by extracting the B₁^- maps. The receive uniformity of the body coil affects the overall image uniformity in situations when the body coil is used as a reference during intensity correction. This also holds true in parallel imaging reconstruction methods using receive arrays in which body coil data are used to normalize the receive array sensitivity profiles, as confirmed in these imaging experiments.⁴

The method proposed here can correct for the receive non-uniformities by means of a single 15 s acquisition. Channel-wise volumetric B₁^- calibration data may also be used for this purpose.⁵ This allows a whole-body protocol to be corrected for the B₁^- without the acquisition of any additional scans.

The DREAM B₁ mapping method can be used for uniformity correction in whole-body applications at 3T. This allows for improving image quality at a minimal burden to the protocol.