Synopsis

Slow acquisitions and susceptibility to respiratory motion artifacts are major challenges in free-breathing 3D whole-heart coronary MR angiography (CMRA). Recently, a respiratory-resolved approach has been proposed to improve scan efficiency and reduce motion artifacts using non-Cartesian acquisitions. However, irregular respirations compromise its suitability for Cartesian imaging. Here, sparsity in a motion-corrected domain is exploited to generate high-quality respiratory-resolved Cartesian images, used to estimate nonrigid motion fields. These are incorporated into a motion-corrected generalized matrix reconstruction, to further improve coronary vessel sharpness. Thus, this approach provides high-quality respiratory-resolved Cartesian CMRA images and a motion-corrected CMRA image at a given respiratory phase.

Introduction

Methods

3D CMRA data is acquired using a prototype golden-step spiral-like Cartesian trajectory,⁸ which samples the k_y-k_z plane with spiral interleaves on a Cartesian grid. Consecutive spirals are separated by the golden-angle. Low-resolution 2D iNAVs are acquired at every heartbeat, before each spiral interleaf of the 3D CMRA acquisition. The 2D iNAVS are used to estimate beat-to beat 2D translational (SI and right-left: RL) motion and obtain the respiratory signal, which is used to distribute the 3D CMRA data into five equally populated respiratory bins. Each undersampled bin is reconstructed using 1) XD-GRASP and 2) XD-ORCCA. The latter is followed by a 3) nonrigid motion-compensated MC-ORCCA reconstruction.

The respiratory-resolved images $\hat{\rm{\mathbf{x}}}_{\rm{b}}$ were obtained by solving: 1) $\hat{\rm{\mathbf{x}}}_{\rm{b}}=\rm\arg\min\limits_{\rm{\mathbf{x}}_b}\left\{ \frac{1}{2}\left\|\mathbf {E}\,\mathbf{x}_b-\mathbf{k}_b\right\|_2^2 +\alpha\,\Psi_t(\mathbf{x}_b)\right\}$ and 2) $\hat{\rm{\mathbf{x}}}_{\rm{b}}=\rm\arg\min\limits_{\rm{\mathbf{x}}_b}\left\{ \frac{1}{2}\left\|\mathbf {E}\,\mathbf{x}_b-\mathbf{d}_b\right\|_2^2 +\alpha\,\Psi_t(\mathbf{x}_b)+\beta\,\Psi_s(\mathbf{x}_b)\right\}$, where $\mathbf{k}_b$ is the binned k-space data, $\mathbf{d}_b$ is the 2D translational corrected binned k-space data, $\Psi_{\rm{s}}$ is the 3D spatial TV function, $\alpha$ and $\beta$ are regularization parameters, $\mathcal{R}\rm{\mathbf{x}}_b=\mathit{T}_{b}\mathbf{x}_b$ is the motion-corrected domain, where $\mathit{T}_{\rm{b}}$ is the translation transform that maps the bin image $\rm{\mathbf{x}}_b$ to the reference image $\rm{\mathbf{x}}_1$ (end-expiration), and $\Psi_{\rm{t}}=\rm{\mathbf{x}}_1-\mathit{T}_{b}\mathbf{x}_b$ is the 1D temporal TV function. The operator $\mathbf{E}=\mathbf{A}_{\rm{b}}\mathbf{FS}$ incorporates the sampling matrix $\bf{A}$ for each bin b, Fourier transform $\bf{F}$ and coils sensitivities $\bf{S}$. The MC-ORCCA reconstruction was obtained by solving 3) $\hat{\rm{\mathbf{x}}}=\rm\arg\min\limits_{\rm{\mathbf{x}}}\left\{ \frac{1}{2}\left\|\mathbf {G}\,\mathbf{x}-\mathbf{d}\right\|_2^2\right\}$, where $\hat{\rm{\mathbf{x}}}$ is the motion corrected image, $\mathbf{G}=\sum\limits_{\rm{b}}\mathbf{A}_{\rm{b}}\mathbf{FS}\mathbf{U}_{\rm{b}}$ and $\mathbf{U}_{\rm{b}}$ are the nonrigid motion fields estimated (via image registration) from XD-ORCCA reconstructed $\hat{\rm{\mathbf{x}}}_{\rm{b}}$ images. These problems were solved with a (non)-linear conjugate gradient method.

In-vivo free-breathing experiments were performed on six healthy subjects and two patients on a 1.5T scanner (Siemens Magnetom Aera) with 18-channel body and 32-channel spine coils. 3D CMRA bSSFP acquisitions were performed using the following parameters: coronal orientation, FOV=320x320x80-104mm³, resolution=1x1x2mm³ (1.2x1.2x1.2mm³, for patients), TR/TE=3.6/1.56ms, flip angle=90°, T₂ preparation (40ms), SPIR-like fat saturation, subject specific mid-diastolic trigger delay, acquisition window ~100ms, 1 spiral interleaf per heartbeat, with acquisition time of 9-12min. For the 2D iNAV acquisition, 14 bSSFP startup echoes were used (same geometry). Patient acquisitions were performed after gadolinium-based contrast injection. Patient 1 had a non-ischemic cardiomyopathy and Patient 2 had known ischemic heart disease with previous stent deployment in the mid-distal segment of the right coronary artery.

Results

Conclusions

Acknowledgements

References

1. Feng L, Axel L, Chandarana H, et al. XD-GRASP: Golden-angle radial MRI with reconstruction of extra motion-state dimension using compressed sensing. MRM 2016; 75:775-788.

2. Piccini D, Feng L, Bonanno G, et al. Four-dimensional respiratory motion-resolved whole heart coronary MR angiography. MRM 2017; 77:1473-1484.

3. Feng L, Chandarana H, Zhao T, et al. Golden-angle sparse liver imaging: radial or Cartesian sampling?. ISMRM 2017; #1285.

4. Correia T, Ginami G, Neji R, et al. Optimized respiratory-resolved motion-compensated 3D Cartesian coronary MRA. ISMRM 2018 (submitted).

5. Henningsson M, Koken P, Stehning S, et al. Whole-heart coronary MR angiography with 2D self-navigated image reconstruction. MRM 2012; 67:437-445.

6. Batchelor P, Atkinson D, Irarrazaval P, et al. Matrix description of general motion correction applied to multishot images. MRM 2005; 54:1273-1280.

7. Cruz G, Atkinson D, Henningsson M, et al. Highly efficient nonrigid motion-corrected 3D whole-heart coronary vessel wall imaging. MRM 2017; 77:1894-1908.

8. Prieto C, Doneva M, Usman M, et al. Highly efficient respiratory motion compensated free-breathing coronary MRA using golden-step Cartesian acquisition. JMRI 2015; 41:738-746.