Introduction

It is challenging to perform single-scan multi-parametric imaging at ultra-high fields (e.g. ≥7T) due to factors such as elevated specific-absorption-rate and peripheral nerve stimulation, longer T₁ and shorter T₂/T₂* relaxometries, increased B₁₊ inhomogeneity and chemical shifts, to name a few. As the result, reports on multi-parametric small animal imaging at ultra-high-fields are rather limited, and all with very limited imaging resolution (e.g. 2D imaging with 0.23mm pixels and 1.5mm slices)^1,2.
Recently, a 3D high-resolution single-scan multi-parametric imaging method, namely MULTIPLEX³, was proposed and demonstrated on 3T. The MULTIPLEX method features in dual-TR, dual-FA and multi-echo GRE acquisition. In principle, GRE-based methods perform better at higher fields in terms of T₁ , PD, susceptibility, SNR and scan times, yet without limitations on resolution and SAR. Therefore, this work aims to perform the initial evaluation of the MULTIPLEX method for ultra-high-resolution imaging on a 9.4T animal system.

Methods

MR Instruments: All MR experiments were performed in a horizontal 30cm-bore 9.4T system (uMR 9.4T, United Imaging Healthcare, Wuhan, China) with gradient performance of 1000mT/m and 10000T/m/s. For ex-vivo experiments, a quadrature volume transceiver coil (24mm inner-diameter) was used. For in-vivo experiments, a quadrature volume transmitter coil (86mm inner-diameter) and a 4-channel rat head surface receiver coil were used.
Ex-vivo rat brain preparations： The ex-vivo rat brain preparation was approved by local IRB, and detailed as previously described⁴. Before all MR scans, ex-vivo brains were placed in Dulbecco's phosphate-buffered saline for at least two days before secured with Fomblin oil.
Animal preparation: With local IRB approval, one healthy adult rat was induced anesthesia using 5% isoflurane and maintained under 1.5% for MRI scan thereafter. Throughout the entire procedure, the rat was monitored (SA Instruments) and well-maintained for rectal temperature at 37°C and respiration rate at ~60bpm.
MR Imaging Parameters: The ex-vivo MULTIPLEX parameters were: coronal FOV=24x16mm², matrix size=208x139x96, voxel size=0.115x0.115x0.120mm³, TR₁/TR₂=7.6/38.4ms, FA₁/FA₂=4°/24°, 5 echoes with TE=2.79~26.99ms, bandwidth=150Hz/px, and TA=10.9hr with NEX=32 and full k-space sampling.
The in-vivo MULTIPLEX parameters were: axial plane with FOV=20x20mm², matrix size=208x208x72, voxel size=0.096x0.096x0.400mm, TR₁/TR₂ = 3/15ms, FA₁/FA₂=4°/16°, 5 echoes with TE=1.10~9.90ms, bandwidth = 450Hz/px, and TA=54min with NEX = 8 and full k-space sampling.
Data Reconstruction: Dicom images of the MULTIPLEX scans were directly reconstructed by both scanners' inline reconstruction pipeline.

Results

Transectional ex-vivo rat brain images are shown in Figure 1, showing typical MULTIPLEX images of PDW, T₁W, aT₁W, SWI and PD/T₁/T₂*/R₂* maps. Figure 2 compares the ex-vivo aT₁W images to a rat brain atlas⁵ to illustrate the fine anatomical details. Figure 3 shows the in-vivo images as well as the scan setups.

Discussion

In this work, we have presented the very first MULTIPLEX results at a 9.4T animal system, showing several interesting findings.
Firstly, both the ex- and in-vivo results served as direct proofs of the MULTIPLEX method on achieving ultra-high-resolution multi-parametric imaging at 9.4T. At such a high field strength, the fast decaying FID signals will limit the SPIRAL and EPI resolution, while the longer tissue T₁ and higher SAR level will respectively increase the scan time of IR- and SE-based method. However, these factors actually worked to the benefit of the GRE-based MULTIPLEX method, where lower FA can be used to achieve proper T₁ weightings and SNR, thus further lowering the already low SAR level and ameliorating B₁ inhomogeneity. Also, faster scans can be achieved by using shorter echo spacing (i.e. higher readout bandwidth) to accommodate the faster T₂* decay. Furthermore, GRE readouts support 3D acquisition for real high resolution imaging, a feature currently not possible on certain multi-parametric methods^6,7.
As illustrated in Figures 1&2, with the 0.12mm isotropic ultra-high-resolution, the ex-vivo images can be viewed from arbitrary cross-sectional planes for direct anatomy delineation with atlas level definition (Figure 2), such as cerebral cortex, corpus callosum, amygdala, hippocampus, olfactory bulb, thalamus, hypothalamus, midbrain, pons, medulla oblongata, spinal cord, cerebellum, caudate putamen, with extension to ventricles and even some other small structures like fornix, optic tract and anterior/posterior commissures.
Secondly, T₁W contrasts of both ex- and in-vivo scans were quite weak (Fig.2) or even reversed (Fig.1), although sufficiently high FA (i.e. 24° and 16°) were already used. This is likely the results of non-T₁ effects like proton density and T₂* relaxation. However, the MULTIPLEX aT₁W and T₁ maps reliably enhanced and quantified the T₁ effects, showing excellent anatomical details³ with expected contrasts.
Thirdly, with the multi-dimensional integration (MDI) method⁸, the resultant aT₁W, T₁ maps and T₂*/R₂* maps from MULTIPLEX are generally spatially uniform, as clearly shown in the in-vivo images . This was the result of removal of coil sensitivity, which was strongest at the top of the head, during MDI calculation⁸.
On the other hand, the residual inhomogeneity in the in-vivo T₁ map suggested the B₁₊ map⁹ requires further tailoring to accommodate the ultra-high-field small FOV scenario. Also the MULTIPLEX protocol should be further optimized for better T₁ contrasts and SNR for in-vivo small animal imaging. Lastly, better receiver coils with round-the-head design can further support acquisition acceleration techniques for practical live animal studies.

Conclusion

In summary, we have demonstrated the ultra-high-resolution imaging capacity of the MULTIPLEX method at a 9.4T animal system, suggesting its potential for single-scan multi-parametric preclinical studies.

Acknowledgements

No acknowledgement found.