Introduction

3D whole brain T1 mapping technique at high magnetic field (B0) poses technical challenges such as inhomogeneous RF transmit (B1+) profile causing inaccurate flip angles in gradient echo sequences. Previous studies focused on low field applications in human brain[1,2] and there is a lack of methodological consideration for high field MRI (9.4T) in spite of its routine use in pre-clinical paramagnetic and manganese contrast enhanced imaging studies. Here we report a simple and accurate 3D T1 mapping technique using a variable flip angles spoiled gradient echo (VFA-FLASH) sequence with B1+ correction implemented at 9.4T. Accuracy of the technique was validated using phantoms and its utility was evaluated in rat brains.

Method

All imaging acquisitions were performed on a Bruker 9.4T/30 MRI instrument equipped with a volume transmit and surface receive coils interfaced with Paravision 5. Phantoms containing various Gd-DOTA (0.02mM~0.27mM) mixed with water were scanned at 37°C. Twelve Wistar rats (body weight: 343±54g; 8~10 weeks old males) were anesthetized with Dexmedetomidine (0.015-0.020 mg/kg/hr) supplemented with 0.5~0.8% isoflurane in a 1:1 Air:O2 mixture and the rats were breathing spontaneously. Animals were positioned supine and normal physiology was assured using a non-invasive vital sign monitoring system. VFA-FLASH technique comprises spoiled gradient echo (3D FLASH TR/TE/FA/TA=15ms/4ms/2,5,10,15,20,30°/4min) taken at multiple flip angles along with double angle B1+ mapping (2D RARE TR/TE/FA/TA=10000ms/22ms/70,140°/10min). Images were reconstructed at 0.24x0.24x0.26mm (128x128x128). In addition, in the phantom study a single slice inversion recovery (IR RARE TR/TE/TI/TA=18000ms/22ms/100~9000ms/8min) sequence was also acquired. T1 maps derived from the VFA-FLASH were calculated using a least square method [3] with B1+ correction, and a three parameters fit was used to derive T1 maps from the IR images. All T1 maps were spatially normalized and population averaged using SPM12 [4] and segmented into grey matter (GM), white matter (WM), and cerebral spinal fluid (CSF) using the Wistar rat brain probability map [5].

Result

In the phantom study, T1s were extracted from region of interests (ROI) in both B1+ corrected and uncorrected VFA-FLASH images, and were plotted against ROIs derived from the IR sequence as shown in Figure. 1. B1+ correction substantially improves the accuracy of the T1 map as evidenced by the fact that discrepancies to the gold standard IR sequence are reduced from ~25% (B1+ uncorrected) to ~6% (B1+ corrected). In regards to the animal studies, representative B1+ and T1 maps are shown in Figure. 2. As expected, spatially dependent B1+ map progressively deviates from a nominal flip angle when the voxel position is further away from the center of the transmit coil, affecting the T1 proportionally. With B1+ correction, as expected a more homogeneous T1 profile is achieved. Spatially normalized and population averaged T1 maps is shown Figure. 2 and T1s were extracted from each tissue compartment as follows: 2121±94ms(GM), 1948±74ms(WM), and 2864±495ms(CSF).

Conclusion

VFA-FLASH with B1+ correction is an accurate and simple 3D T1 mapping technique that can readily be implemented in brain as well as other body organs. The phantom study demonstrated that a wide range of T1s (1000~3300ms) were consistent with the gold standard IR sequence; further, whole brain GM, WM, and CSF T1s were also comparable to T1s reported in previous studies [6]. Despite the use of a whole body volume coil, substantial B1+ inhomogeneity within the brain was still observed and must be corrected for at 9.4T in order to achieve accurate T1 measurements.

Acknowledgements

This study is supported by NIH-R01AG048769 and NIH-RF1AG053991.

References

1] Deoni et al. MRM (2003) [2] Marques et al. Neuroimage (2010) [3] Wang et al. MRM (1987) [4] Ashburner et al. Neuroimage (2005) [5] Valdes PA et al. Front in Neuroinfo (2011) [6] de Graaf et al. MRM (2006)