Introduction

Quantitative MRI isolates tissue properties from the influence of MR parameters and aid in earlier disease detection¹, longitudinal studies², and monitoring treatment³. Moreover, a good quantitative map depends mildly on protocols and scanners and can be used for machine-learning diagnosis without retraining for different protocol and hardware.
The gold standard methods of T1 and T2 mapping methods require long acquisition times. For clinical applications, several accelerated methods were proposed. MOLLI acquires several echoes during magnetization recovery and T1 maps can be calculated based on data collected in a single breath hold⁴. GRAPPATINNI accelerates multiple spin-echo acquisitions by reducing spatial encodings and reconstructs the T2 map using a model-based method⁵. However, the quantification is usually biased in the 2D methods due to the imperfect slice profile and B₁ inhomogeneity. 3D mapping methods can reduce bias by 3D excitation and using actual flip angles in T1 and T2 estimates. By acquiring spoiled gradient echo (SPGR) images at two different flip angles, we can estimate T1 maps⁶. Two images from large flip angle dual-echo steady-state (DESS) can be combined to approximate a T2 map⁷. However, the signal equation for T1 estimation depends slightly on T2 values and vice versa. Here, we propose a new simultaneous T1 and T2 mapping method, which uses 3D excitation and actual flip angles in quantification. Moreover, a magnetization transfer (MT) weighted image will be simultaneously calculated without extra scan time.

Methods

The pulse sequence of multi-echo DESS was shown in Figure 1. In this study, we only implemented a single-echo version of it. It’s crucial that TE₁ equals TE₁' such to eliminate the effect of T2’. The signals of SPGR and DESS were modeled based upon the extended phase graph theory⁸:
$$S_{SPGR}(α,TR_{SPGR},TE)=M_{0,SPGR}×f_{EPG} (T_1,T_2,TR_{SPGR},α,Ψ)×\exp(-\frac{TE}{T2}-\frac{TE}{T2'}),$$
$$S_{DESS,0} (α,TR_{DESS},TE_n )=M_{0,DESS}× \tan⁡(\alpha/2)⋅(1-(E_1-\cos⁡\alpha )⋅r)⋅\exp⁡(-\frac{TE_n}{T2}-\frac{TE_n}{T2' }),$$
$$S_{DESS,-1} (α,TR_{DESS},TE_m' )=M_{0,DESS}×\tan⁡(\alpha/2)⋅(1-(1-E_1 \cos\alpha )⋅r)/E_2 ⋅\exp⁡(-\frac{TR{DESS}-TE_m'}{T2}-\frac{TE_m'}{T2' }),$$
$$E_{1,2}=\exp⁡(-\frac{TR_DESS}{T_{1,2}}),$$
$$p=1-E_1\cos⁡\alpha-E_2^2(E_1-\cos⁡\alpha⁡),$$
$$q=E_2(1-E_1)(1+\cos⁡\alpha),$$
$$r=(1-E_2^2) (p^2-q^2)^{-1/2}$$
, where f_EPG is the F₀ at steady-state and was calculated using numerical methods.
Ψ is the RF-pulse phase-difference increment for RF spoiling. By choosing the same TE and TE’ in SPGR and DESS, we can solve for the unknowns, T1, T2, $$$M_{0,SPGR}, M_{0,DESS}$$$ from four signal equations of $$$S_{SPGR} (\alpha_1,TR_{SPGR},TE),S_{SPGR} (\alpha_2,TR_{SPGR},TE),S_{DESS,0} (\alpha_3,TR_{DESS},TE),S_{DESS,-1} (\alpha_3,TR_{DESS},TE)$$$. Due to the large RF power difference between SPGR and DESS, $$$M_{0,SPGR}, M_{0,DESS}$$$ can be used to calculate MT weighted images: $$$MTw=(M_{0,SPGR}-M_{0,DESS})/M_{0,SPGR}$$$.
We acquired four sets of images of the knee using the SPGR (TE=3ms,TR=10ms,flip angle=4^o/21^o, FOV=160x160x56mm³, and matrix size=256x256x28) and the DESS (TE=3ms,TR=20ms, flip angle=40^o, FOV=160x160x56mm³, and matrix size=256x256x28) sequences. We used the same protocol for the phantom studies apart from using a different matrix size (160x160x28). Flip angle maps were acquired using pre-saturation turbo fast low angle shot (FOV=200x200x87.5mm³, and matrix size=64x64x28). The flip angle maps were used to estimate T1 and T2. A phantom was built, consisting of four tubes to evaluate the accuracy of our method. Each tube contains 2.27%, 2.86%, 3.43%, and 4% of agarose respectively and 0.003M of CUSO₄.
The gold standard T1 and T2 values of these four tubes were determined separately by a 1D spin-echo sequence (TE=10/30/50/70/90/110/130/150ms, and TR=20000ms) and inversion recovery spin-echo sequence (IR=200/400/600/800/1000/1200/1400/1600/1800/2000ms, TE=10ms, and TR=20000ms). We compared our method with three well-known T1 and T2 mapping techniques: the conventional B₁ corrected SPGR with two flip angles for T1 mapping, DESS T2 mapping using the same data acquired as for our method, and the multi-echo spin-echo (MESE) method (TE= 13.8/27.6/41.4/55.2/69ms, TR=1980ms, FOV=128x128x56mm³, and matrix size=128x128x28) for T2 mapping. A healthy volunteer and the phantom were scanned. All experiments were conducted at a 3T MR scanner (MAGNETOM Prisma, Siemens, Erlangen, Germany).

Results

Figure 2AD shows the T1 and T2 maps using the proposed method. The T2 values are lower compared with those acquired using the MESE method (Figure 2C). Visually, T1 and T2 maps are similar to those reconstructed using SPGR and DESS separately (Figure 2). Table 1 compared the mean T1 and T2 values estimated by different methods. The T2 values estimated by the proposed method deviate less than 1% from the gold standard, whereas MESE presents a strong bias of T2 values. The T2 values were underestimated using DESS alone and the bias increases when T2 is higher. For T1 maps, the proposed method is more accurate than the SPGR method with 2% bias compared with the gold standard. Figure 3A and 3B shows the M₀ images estimated for SPGR and DESS sequences. M_0,SPGR has more homogeneous contrast, which is similar to the PD contrast image but M_0,DESS presents hypo-intensities in muscle and cartilage. By combining these two images, an MT-weighted image (Figure 3C) was reconstructed. Figure 4 shows the T1 and T2 maps calculated by the proposed method.

Discussion and Conclusion

We proposed a method to simultaneous acquire T1, T2, PD weighted, and MT weighted images. The scan time is the same as conventional SPGR T1 and DESS T2 mapping, but with an accuracy comparable to SE T2 and IR-SE T1 mapping. In the future, we will extend this approach to a multi-echo version and acquire T2’ map as well without extra scan time. Moreover, we can replace the SPGR sequence with multi-echo SPGR sequence or multi-echo DESS sequence to acquire more information for higher image quality.

Acknowledgements

No acknowledgement found.