Synopsis

Whole-heart sub-millimeter isotropic coronary magnetic resonance angiography (CMRA) provides detailed information of the coronary arteries and surrounding vessels. Recently, a patch-based reconstruction technique (3D PROST) has been proposed to achieve sub-millimeter isotropic resolution CMRA in a predictable scan time. However, this approach only corrects for 2D translational respiratory motion of the heart and image quality can be affected by residual non-rigid motion. Here we propose to integrate 3D PROST into a highly accelerated non-rigid motion correction framework to achieve high quality whole-heart free-breathing isotropic sub-millimeter Cartesian CMRA in a clinically feasible scan time. The feasibility of the proposed method was tested in seven healthy subjects and two patients with suspected coronary artery disease.

Purpose

Methods

Acquisition & Reconstruction – Undersampled acquisitions are performed with a variable density Cartesian acquisition with spiral-like order [2,5]. A 2D iNAV precedes each spiral arm acquisition to enable beat-to-beat 2D translational respiratory motion correction without any data rejection. To account for 3D non-rigid motion, respiratory binning is performed by sorting the CMRA data into five respiratory phases. High-quality respiratory-resolved images are reconstructed using the recently proposed XD-ORCCA technique [6] and used to estimate 3D bin-to-bin non-rigid motion fields [7]. A single-phase motion-compensated 3D CMRA image is then reconstructed by integrating the obtained non-rigid motion fields into 3D-PROST reconstruction (Figure 1) which aims at solving the following non-rigid-PROST problem:

$$\mathcal{L}\left(x,\mathcal{T} \right) := \underset{x,\mathcal{T}}{\operatorname{argmin}} \frac{1}{2}\Vert Ex-y \Vert_2^2 + \lambda\sum_p \Vert \mathcal{T}_p \Vert_\ast \quad s.t. \quad \mathcal{T}_p = R_p \left( x \right) \quad \quad \quad \quad [1]$$

Where $E=\sum_{b=1}^5A_bFS_cU_b$ is the encoding operator (including coil sensitivity maps $S$, Fourier operator $F$ and sampling $A$), $U_b$ are the 3D non-rigid spatial transformations for motion state $b$, $y$ denotes the 2D translationally corrected undersampled data and $x$ is the image to reconstruct. The operator $R_p \left( . \right)$ constructs a matrix of non-local similar 3D patches from the patch $p$ centered at pixel $p$. The nuclear norm is used to enforce low-rank on a patch scale and $\lambda>0$ controls the strength of sparsity.

Optimization – Equation 1 can be solved using the alternating direction method of multipliers (ADMM) which consists of splitting the main optimization problem $\mathcal{L}$ into two simpler sub-problems: 1) a parallel imaging regularized motion-compensated MR reconstruction (optimization on $x$, solved with conjugate gradient optimization) [4], and 2) a 3D patch-based denoising (optimization on $\mathcal{T}$, solved by singular value thresholding) [2]. The following parameters were empirically selected to provide the best reconstruction quality: patch size=5x5x5 voxels, search window=21x21x21 voxels, conjugate gradient iterations=5, $\lambda=0.1$, number of similar patches=10, patch offset=4, ADMM penalty parameter=0.3, ADMM iterations=6.

Imaging – Seven healthy subjects (4 males, 32±9 years) underwent whole-heart free-breathing CMRA on a 1.5T scanner (Siemens Magnetom Aera). Data were acquired without contrast agent administration with the following parameters: ECG-triggered 3D bSSFP sequence, 0.9mm³ isotropic resolution, undersampling factor of 5, FOV=320x320x86-115mm³, FA=90°, T2-preparation duration=40ms, TE/TR=1.6/3.7ms, bandwidth=890Hz/pixel, subject-specific mid-diastolic acquisition window (range ~92-118ms). Images were reconstructed to a resolution of 0.6mm³ and vessel sharpness and length of the right and left coronary arteries (RCA/LAD) were measured after reformatting [8]. In addition, acquisitions were performed in two patients with suspected coronary artery disease with the same parameters as in the healthy subjects study but with undersampling factors of 3 and 4 respectively. Reformatted images from patients were compared to conventional CT coronary angiography (CTCA).

Results

Conclusion

Acknowledgements

References

[1] Akçakaya M, Basha TA, Chan RH, et al. Accelerated Isotropic sub-millimeter whole-heart coronary MRI: compressed sensing versus parallel imaging, Magn Reson Med. 2014;71:815-822.

[2] Bustin A, Ginami G, Cruz G, et al. Five-minute whole-heart coronary MRA with sub-millimeter isotropic resolution, 100% respiratory scan efficiency, and 3D-PROST reconstruction. Magn Reson Med. 2018;00:1-14.

[3] Henningsson M, Koken P, Stehning C, et al. Whole-heart coronary MR angiography with 2D self-navigated image reconstruction. Magn Reson Med. 2012;67:437-445.

[4] Cruz G, Atkinson D, Henningsson M, et al. Highly efficient nonrigid motion-corrected 3D whole-heart coronary vessel wall imaging. Magn Reson Med. 2016;77:1894-1908.

[5] Prieto C, Doneva M, Usman M, et al. Highly efficient respiratory motion compensated free-breathing coronary MRA using golden-step cartesian acquisition. J. Magn. Reson. Imaging 2015;41:738-746.

[6] Correia T, Ginami G, Cruz G, et al. Optimized respiratory-resolved motion-compensated 3D Cartesian coronary MR angiography. Magn Reson Med. 2018;00:1-12.

[7] Modat M, Ridgway G, Taylor Z, et al. Fast free-form deformation using graphics processing units. Comput Meth Prog Bio 2010;98:278-284.

[8] Etienne A, Botnar RM, Van Muiswinkel AM, et al. Soap-Bubble“ visualization and quantitative analysis of 3D coronary magnetic resonance angiograms. Magn Reson Med. 2002;48:658-666.