Introduction

It has long been a challenge for most currently available multi-parametric imaging methods to achieve satisfactory performance on ultra-high fields (>3T) due to several major factors, such as the elevated specific absorption rate (SAR) and peripheral nerve stimulation (PNS) level, longer T₁ and shorter T₂/T₂* relaxometries, stronger B₁₊ inhomogeneity and chemical shifts, to name a few. For instance, the increased SAR level prohibits faster scans with spin echo based methods¹, the longer T₁ requires longer TR and/or TI (thus longer scan times) for adequate signal recovery¹, and the shorter T₂* effectively reduces the imaging resolution of EPI² and SPIRAL³ methods through the intrinsic filtering effect during readouts.
Recently, a dual-TR, dual-FA, multi-echo GRE based multi-parametric imaging method, namely MULTIPLEX⁴, was proposed and demonstrated on 3T. In principle, GRE based methods can offer better performance at ultra-high fields in terms of T₁, PD, susceptibility, SNR and scan times, while without limitations on imaging resolution or SAR levels. However, the more severe B₁₊ field inhomogeneity may potentially hamper the accuracy of T₁ mapping⁵. Therefore, this work aims to evaluate the MULTIPLEX performance at a newly introduced field strength of 5T⁶, in comparison with well-controlled 3T results.

Methods

High resolution MULTIPLEX brain images were collected on the same healthy volunteer (male, 36/yro, passed safety screening and gave written consents) on a 5T whole body scanner (uMR Jupiter, United Imaging Healthcare, Shanghai, China) with a 48-channel head coil, as well as on a 3T scanner (uMR790, United Imaging Healthcare, Shanghai, China) also with a 48-channel head coil.
For this initial comparison tests, the same scanning protocol was used on both fields, including following key parameters⁴: 7 echoes with TE = 3.05~23.24ms, TR₁/TR₂ = 7.54/32.46ms, FA₁/FA₂ = 4°/16°, matrix size = 272x231x72 and voxel size = 0.8x0.8x2mm³, bandwidth = 260Hz/px. With 2x acceleration, the total scan time was 8:10.
Dicom images of the MULTIPLEX scans were directly reconstructed by the integrated reconstruction pipeline on both scanners and can be readily compared. Furthermore, raw k-space data of both scans were saved and processed offline to evaluate the imaging signal-to-noise ratio (SNR) of individual echo images.

Results

The MULTIPLEX B₁₊ and T₁ maps are shown in Figure 1. Other typical MULTIPLEX images are compared in Figure 2, including augmented T₁W (aT₁W) images, PD/R₂*/QSM maps and SWI mIP images. ROI values from typical tissues are shown and compared in Table 1. And the image magnitude, noise distribution and SNR maps of the same MULTIPLEX echo (i.e. FA₁-TR₁-Echo₁) from both fields are compared in Figure 3.

Discussion

In this work, we have presented the initial experience of MULTIPLEX scans on a state-of-the-art 5T whole body scanner, which were compared to 3T results.
The 5T field strength has only recently become available for whole body imaging⁶, thus no literatures on the relaxometry properties (e.g. T₁/T₂*) or other quantitative quantities (e.g. PD/QSM) are readily available. However, referencing the 3T and 7T literatures⁷, one can safely assume that these MRI signal properties at 5T will likely fall between 3T and 7T. By using an identical scanning protocol, reliable and similar outcomes on both 5T and 3T should be expected.
The major high-field challenge in MULTIPLEX is expected to be T₁ mapping with B₁₊ correction, like any other dual-FA methods⁵. With the dual-TR design and actual flip angle imaging (AFI) technique⁸, the B1 fields can be reliably extracted at both fields. As shown in Figure 1, the B₁ field inhomogeneity at 5T is indeed more severe than at 3T due to stronger dielectric effects⁹. Nevertheless, the 5T T₁ map became homogeneous (Fig.1) and yielded reasonable values (Table 1) after B₁₊ correction. However, some unevenness in 5T T₁ maps can still be observed, most likely due to the higher order components in the B₁₊ fields not being taken account for, and it will be our future focus to further improve the B₁₊ correction effect.
Table 1 also indicates that T₂* values at 5T were lower than at 3T, which was expected. Proton density (or relative PD to be accurate) and QSM values, on the other hand, remained relatively consistent between the two fields, as in principle both quantities are field independent. And with the same echo times, SWI at 5T showed stronger susceptibility weighted contrasts than its 3T counterpart (Fig.2). These observations suggest the 5T MULTIPLEX results were generally better than 3T's, while well within reasonable and expected range.
Generally, 5T results yielded better imaging quality and higher SNR than 3T (Figs.1~3). This is clearly shown in Figure 3 that 5T images had slightly higher signal intensity, lower noise level and subsequently higher SNR. For future works, the MULTIPLEX protocol can be and should be optimized specifically for 5T or even higher fields like 7T to achieve further optimized outcomes, such as higher resolution and/or shorter scanning time. Also, more stringent tests are needed to validate the accuracy of MULTIPLEX's quantitative maps at 5T.

Conclusion

In conclusion, we have for the first time demonstrated the feasibility of MULTIPLEX imaging on a 5T whole body scanner, suggesting a strong potential of MULTIPLEX as a high field multi-parametric imaging solution.

Acknowledgements

No acknowledgement found.