2215

Off-resonance error suppression in MOLLI T1 mapping using a specific CE-FAST
Mostafa Berangi1, Vahid Ghodrati2,3, Fereshteh Hasanzadeh4, Puria Rafsanjani1, Zahra Alizade Sani5, and Abbas Nasiraei Moghaddam1

1Faculty of Biomedical enginnering, AmirKabir University of technology, Tehran, Iran (Islamic Republic of), 2Department of Radiological Sciences, University of California, Los Angeles, CA, United States, 3Biomedical Physics Interdepartmental Program, Los Angeles, CA, United States, 4School of Allied Medical Sciences, Shahid Beheshti University of Medical Sciences and Health Services, Tehran, Iran (Islamic Republic of), 5Rajaei Cardiovascular Medical and Research Center, Iran University of Medical Sciences, Tehran, Iran (Islamic Republic of)

Synopsis

MOLLI is a well-known method for cardiac T1 mapping. Data sampling in MOLLI is based on SSFP readout. It is therefore susceptible to off-resonance which is considered an error source for T1 mapping. We have presented a method to suppress off-resonance variation error in T1 estimation by means of a specific CE-FAST in combination with MOLLI. We have evaluated our method in simulation and Phantom experiment with promising results.

INTRODUCTION

Off-resonance rotation, which is the result of B0 field in-homogeneity, is an inevitable phenomena in MR imaging. Balanced Steady State Free Precession (b-SSFP) sequence is in particular sensitives to off-resonance, resulting in well-known dark-band artefact. MOLLI1 as a clinically available sequence for cardiac T1 mapping consists of several b-SSFP image acquisitions and thus T1 estimation using MOLLI is sensitive to off-resonance, even in regions without dark band 2.

We have proposed a new method based on a specific CE-FAST3 acquisition instead of the b-SSFP aiming to suppress the effect of both small and severe off-resonance values which result in T1 estimation error and dark band artefact respectively. The feasibility of decreasing the off-resonance error through this method and its robustness has been investigated and quantified in this study in simulation and a phantom experiment.

METHODS

The b-SSFP signal amplitude has a periodic behavior over extra-acquired phase per TR with a period of 2π as is shown in Fig.1a. Following a method, implemented for SSFP-fMRI4, here we proposed a technique by means of a specific implementation of CE-FAST (in which an extra gradient field in reading direction is applied) an exact 2π off-resonance across each voxel in every TR is induced. Therefore one period of off-resonance profile is squeezed in each voxel (Fig.1b) and the resulting signal of each voxel is the average of entire b-SSFP profile. This averaged signal is therefore, independent to voxel position and local field strength variation.

Proposed method is simulated using numerical solution of Bloch equation for a single voxel to assess the performance of T1 estimation. To evaluate the feasibility of proposed method in phantom experiment, series of 5(3)3 MOLLI acquisitions with different values of frequency offset were collected in a SIEMENSE Aera scanner (with parameters: TE=1.2ms, TI 1&2 (for each LL experiment) = 100 ms & 180 ms and FA=35°). Acquired data consists of different MOLLI datasets with 5Hz resolution in off-resonance frequency over one b-SSFP profile period for a fixed array of phantoms. Then the CE-FAST signal was modeled using Eq.1.

$$S_{CE-FAST}=\frac{1}{N}\sum_{i=1}^NS_i^{MOLLI} \qquad Eq.1$$

Where $$$S_i^{MOLLI}$$$ is the complex signal of each MOLLI dataset corresponding to each voxel which is acquired with variable frequency offset and N is the total number of MOLLI datasets with variable frequency offset. Then $$$S_{CE-FAST}$$$ is further used for T1 estimation.

RESULTS

Fig.2-A represents the simulation of one LL (Look-Locker) experiment of MOLLI dataset for several off-resonance values from 0° to 180°(each colored curve) and Fig.2-B is the CE-FAST signal of same experiment as a function of time. Fig.3 shows two T1 maps of phantom experiments generated using same dataset. Each map contains 4 different phantoms (left to right, phantom1 to 4) with different T1 values. The left map corresponds to original MOLLI and the right map is generated using proposed method. Also Table-1 corresponds to mean & Std. of estimated T1 values of phantoms.

DISCUSSION

As shown in Fig.2 even small values of off-resonance result in error in T1 estimation using MOLLI. Proposed method by replacing b-SSFP acquisition with CE-FAST, makes T1 estimation robust to off-resonance. With respect to Fig.3 and Table-1, T1 estimation of different uniform phantoms has much smaller variation compared to those of original MOLLI. Also for phantom#1 in T1 map of original MOLLI, there are two bars with under estimation for T1 which is the result of local field inhomogeneity while in proposed method this off-resonance related error is eliminated and almost uniform estimation is achieved. Therefore two advantages of proposed method are: 1) Removing the variations of T1 estimation caused by off-resonance in the small off-resonance frequencies (pass-band of SSFP profile). 2) Suppression of SSFP profile stop-band error (dark band artefact) in T1 estimation.

CONCLUSION

Using CE-FAST in combination with MOLLI has potential to suppress off-resonance variation error of T1. In MOLLI there is a maximum error of -10% and its error is related to local off-resonance amount (and other imaging parameters) while in CE-FAST-MOLLI error is independent to local off-resonance.

Acknowledgements

No acknowledgement found.

References

1. Daniel R. Messroghli, Aleksandra Radjenovic, Sebastian Kozerke, David M. Higgins, Mohan U. Sivananthan and John P. Ridgway. Modified Look-Locker Inversion Recovery (MOLLI) for High-Resolution T1 Mapping of the Heart. Magn Reson Med 52:141–146 (2004)

2. Kellman, Peter, et al. "Influence of Off-resonance in myocardial T1-mapping using SSFP based MOLLI method." Journal of Cardiovascular Magnetic Resonance 15.1 (2013): 63.

3. M.L. Gyngell. The application of steady-state free precession in rapid 2DFT NMR imaging: FAST and CE-FAST sequences. Magn. Reson. Imaging. 6 (1988) 415–419.

4. V. Malekian, A. Nasiraei, M. Khajehim, D. G. Norris. A Robust SSFP Technique for fMRI at Ultra-High Field Strengths. Proc. Intl Soc Mag Reson Med. Vol. 25. 2017.

5. Mahdi Khakeim (2016), Improving reproducibility and robustness of fMRI in Averaged-BOSS method through GRE BOLD Inspired Processing Techniques,(Unpublished Master’s dissertation), , AmirKabir University of Technology.

Figures

Figure 1-a) periodic behavior of b-SSFP signal over off-resonance frequency with period of 2π. B) off-resonance profile is squeezed to each voxel using an extra gradient, thus each voxel signal is an average of all off-resonance related signals 5.

Figure 2- A) simulation of (normalized) longitudinal magnetization vector over time in a Look-Locker (LL) experiment of MOLLI dataset for a range of 0° to 180° off-resonance variation (each colored curve) representing off-resonance sensitivity of MOLLI signal which causes error in T1 estimation. B) Simulation of (normalized) longitudinal magnetization vector over time for CE-FAST signal. As is shown, this signal is robust to off-resonance.

Figure 3- T1 map of 4-uniform phantoms with different T1 values. Left map is calculated using original MOLLI and right map is calculated using proposed method.

Table 1--Mean and std. of T1 values in Fig.3 (phantoms from right to left are numbered 1 to 4). As is shown both maps have almost same mean but std. in original MOLLI is much more than proposed method.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2215