Synopsis

Long acquisition window in myocardial T₁ mapping degrades the image quality. Physiological motion induced fluctuations along the parametric direction corrupt the similarity among T₁-weighted images and adversely impact the performance of the compressed sensing-based methods. In this study, we propose a two-layer sparsifying transform (MO₂) combined with motion correction to shorten the acquisition window of myocardial T₁ mapping. Results showed that the proposed method enabled higher acceleration factor. The resulting shorter acquisition window and fewer RF saturation pulses increased the accuracy of the native T₁ measurement.

Introduction

Methods

The reconstruction step are summarized in Figure 1. First, T₁-weighted images reconstructed using zero filling method are used for rigid registration. The estimated global translational motion is used for k-space data correction using Fourier shift theorem. The corrected k-space data are reconstructed using a two-layer sparsifying transform. Specifically, principal component analysis (PCA) is applied along parametric direction and then dictionary learning is used to reconstruct the sparse PCA coefficients. The T₁-weighted images are obtained by transforming the resulting PCA coefficients with inverse PCA transform. The reconstruction problem can be formulated as

$$minimize_{I_p,\alpha_n}\sum_n||D\alpha_n-R_n(B[I_1,...,I_{N_p}])||_2^2+\upsilon\sum_{c,p}||k_{c,p}-F_pS_cI_p||_2^2$$

$$subject\ \ to\ \forall{n}, ||\alpha_0||_0\leq{T_0}$$

where I_p is the p-th vectorized T₁-weighted image to be reconstructed, with p = 1, …, N_p. N_p is the total frame number. F_p is the undersampled Fourier operator. S_c is the c-th coil sensitivity and k_c,p is the undersampled k-space data.$$$B=\left\{\overrightarrow{b_1},\overrightarrow{b_2},...,\overrightarrow{b_L}\right\}$$$is the matrix of the PC basis, which is generated using a training data set from T₁-weighted images with the singular value decomposition algorithm. R_n represents the operator that extracts the PC coefficients at the same spatial location. D is the patch based dictionary. α_n is the sparse representation of patch n with respect to D. T₀ is the required sparsity level. ν is the regularization parameter.

Evaluation:

All MR data were acquired on a 1.5 T Philips Achieva (Philips Healthcare, Best, The Netherlands) system with a 32-channel cardiac coil. A random undersampled k-space data acquisition scheme at R=2, 3, and 4 (Figure 2) was implemented for slice-interleaved myocardial T₁ mapping sequence (STONE) [5] to allow prospectively randomly undersampling data. T₁MES Phantom [6] were scanned with the above acquisition scheme to assess the accuracy and precision of T₁ measurements related to the reconstruction method. Imaging parameters were: bSSFP readout with TR/TE/flip angle=3.12ms/1.56ms/70˚, field-of-view=240×228mm2, matrix size=176×167, and simulated ECG of 60 bpm. Reference T₁ values of the phantom were measured using an inversion recovery spin echo sequence. For in-vivo experiments, three healthy volunteers (two females, 24±3 years old) were recruited for this study (IRB proved, written informed consents obtained). In each subject, we acquired 6 datasets using prospective undersampling at R=2 through 4 with both random and uniform undersampling masks using STONE sequence. Imaging parameters were the same as in phantom study except TR/TE=2.8ms/1.4ms and FOV=360×342mm². The proposed method was used to reconstruct T₁-weighted images with random undersampling masks and SENSE was used for uniform undersampling reconstruction. All T₁ maps were calculated using a two-parameter model from the T₁-weighted images.

Results

Conclusion

Acknowledgements

References

1. Messroghli DR, Moon JC, Ferreira VM, Grosse-Wortmann L, He T, Kellman P, Mascherbauer J, Nezafat R, Salerno M, Schelbert EB, Taylor AJ, Thompson R, Ugander M, van Heeswijk RB, Friedrich MG. Clinical recommendations for cardiovascular magnetic resonance mapping of T₁, T₂, T₂* and extracellular volume: A consensus statement by the Society for Cardiovascular Magnetic Resonance (SCMR) endorsed by the European Association for Cardiovascular Imaging (EACVI). J Cardiovasc Magn Reson 2017;19(1):75.
2. Kang J JJ, Tarokh V, Nezafat R. Dictionary-based Reconstruction for Free-Breathing Myocardial T₁ Mapping. 2017; Honolulu,Hi, USA. In Proceedings 23th Scientific Meeting, International Society of Magnetic Resonance in Medicine (ISMRM). p 545.
3. Li W, Griswold M, Yu X. Fast cardiac T₁ mapping in mice using a model-based compressed sensing method. Magn Reson Med 2012;68(4):1127-1134.
4. Zhu Y, Zhang Q, Liu Q, Wang YX, Liu X, Zheng H, Liang D, Yuan J. PANDA-T1rho: Integrating principal component analysis and dictionary learning for fast T_1ρ mapping. Magn Reson Med 2015;73(1):263-272.
5. Weingärtner S, Roujol S, Akçakaya M, Basha TA, Nezafat R. Free‐breathing multislice native myocardial T₁ mapping using the slice‐interleaved T₁ (STONE) sequence. Magn Reson Med 2015;74(1):115-124.
6. Captur G, Gatehouse P, Keenan KE, Heslinga FG, Bruehl R, Prothmann M, Graves MJ, Eames RJ, Torlasco C, Benedetti G. A medical device-grade T1 and ECV phantom for global T1 mapping quality assurance—the T₁ Mapping and ECV Standardization in cardiovascular magnetic resonance (T1MES) program. J Cardiovasc Magn Reson 2016;18(1):58.