2053

Free-breathing whole-heart coronary MRA with Radial SSFP and fully automated 3D rigid body motion corrected reconstruction
Guruprasad Krishnamoorthy1,2, Joao Tourais1,2, Jouke Smink1, Marc Kouwenhoven1, Suthambhara Nagaraj3, and Marcel Breeuwer1

1MR Clinical Science, Philips Healthcare, Best, Netherlands, 2Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands, 3MR R&D, Philips Healthcare, Bangalore, India

Synopsis

Prolonged acquisition time and susceptibility to respiratory motion remain to be the major challenge in 3D free-breathing whole-heart MR angiography. Recently, respiratory self-navigated 3D-Radial CMRA technique has been proposed to address the said limitations. In this technique, a 1-D projection oriented in Superior-Inferior direction are acquired at every heartbeat before each interleave of 3D-Radial imaging data for 1D motion correction of the heart with 100% respiratory gating efficiency. In this work, an extension of this method is proposed to estimate and correct for the 3D motion of the heart along with a robust navigator processing and automatic coil selection algorithm (3D-PRONAV). The proposed method is validated in five volunteers.

Introduction

Prolonged acquisition time and susceptibility to respiratory motion remain to be the major challenge in 3D free-breathing whole-heart (WH) MR angiography (CMRA). Cartesian acquisition in combination with respiratory gating and tracking is still the most widely used technique [1]. The relatively small acceptance window and the requirement of careful planning of the Navigator beam limit its applicability in clinical practice. Recently, respiratory self-navigated 3D-Radial CMRA technique has been proposed to address the above-mentioned limitations [2]. In this technique, a 1-D projection oriented in Superior-Inferior (SI) direction is acquired at every heartbeat before each segment of 3D-Radial data acquisition and is used for estimation of the rigid-body motion of the heart, allowing motion correction with100% respiratory gating efficiency. However, this method does not account for the motion of the heart in Anterior-Posterior (AP) and Right-Left (RL) directions which can lead to a local blurring of images [3]. In this work, an extension of this method is proposed to estimate the 3D motion of the heart with a robust navigator processing and automatic coil selection algorithm (3D-PRONAV). The proposed method is validated in five volunteers.

Methods

The proposed acquisition scheme is based on a balanced-SSFP sequence with 3D-Radial trajectory based on Spiral Phyllotaxis [4] modified to support anisotropic field-of-views (FOV) in XY and Z directions [5]. Three orthogonal projections were acquired in AP, RL and SI orientations (3D-PRONAV) respectively before each segment of image acquisition to derive motion information. The sequence is shown in figure 1. The algorithm proposed in this work to derive motion information from the 3D-PRONAV has the following steps,

1. Segmentation of the blood pool (described in [2]) in the reference projection followed by cross-correlation to estimate motion with sub-mm precision from each anterior coil elements, mi - the motion estimate from coil i with i = 1,2.. N coil elements.

2. Computing motion based on center-of-mass in the quadrature body coil (COMqbc ) [6].

3. Selecting a coil which has the maximum correlation coefficient (ρmax), where ρmax = max(|ρ(COMqbc, mi )|), i Є {1,2,..N}, and ρ(COMqbc, mi) is the correlation coefficient (range from -1 to 1) between COMqbc and mi.

These steps were repeated for all the three projections to estimate the 3D rigid-body motion of the heart. 3D motion correction is performed in k-space prior to gridding according to the Fourier shift theorem. The proposed technique is implemented in a 1.5T Ingenia scanner (Philips Healthcare, Best, Netherlands). Five healthy volunteers were imaged for free-breathing WH CMRA. The images were analyzed using Soap-Bubble tool [7]. A pencil beam navigator motion data was additionally acquired for data analysis. The imaging parameters were; TE/TR: 1.7 / 3.7 ms, acquired aliasing-free FOV: ~ 117 X 117 X 58 mm, reconstructed FOV: (220 mm)3, resolution: (1.2 mm)3, flip angle: 90⁰, receiver bandwidth: 867 Hz/Pixels, volume selective with an FWHM of 220 mm in SI direction was used, 2 saturation slabs were positioned to suppress signals from right and left shoulders respectively, a total of 9048 to 12064 radial projections were acquired depending on the volunteer’s heart rate in 377 interleaves. The scan time for the proposed technique was in the range of 4.5 to 7.2 mins depending on the volunteer’s heart rate. Reconstructions were performed on the scanner immediately after acquisition with the reconstruction times of approx. 30sec.

Results

An example of the proposed motion estimation from the 3D PRO-NAV is shown in figure 2. Blood-pool segmentation in reference projections of 3D-PRONAV is shown in 2(a) and the estimated motion are shown in 2(b). The COMqbc is plotted in 2(b) and selected coils in 2(b) are highlighted. Comparison between the motion derived in SI from 3D-PRONAV and the pencil beam navigator is shown in Figure 3. Free-breathing WH CMRA images obtained from one of the volunteers using the proposed technique are shown in figure 4. The measured length of major coronary branches in five volunteers is shown in figure 5.

Discussion and Conclusion

A new fully automatic method to estimate 3D translation motion of the heart from projection navigators (3D-PRONAV) has been proposed and validated. Motion correction in 3D resulted in better visualization of coronary arteries and overall sharper cardiac images compared to motion correction performed only in SI orientation for scans with 100% respiratory gating efficiency

Acknowledgements

This work was supported by the European Commission within the Horizon 2020 Framework through the MSCA-ITN-ETN European Training Networks (project number 642458).

References

1. Stuber, M., et al., Submillimeter Three-dimensional Coronary MR Angiography with Real-time Navigator Correction: Comparison of Navigator Locations. Radiology, 1999. 212(2): p. 579-587.

2. Piccini, D., et al., Respiratory self-navigation for whole-heart bright-blood coronary MRI: Methods for robust isolation and automatic segmentation of the blood pool. Magnetic Resonance in Medicine, 2012. 68(2): p. 571-579.

3. Lai, P., et al., A respiratory self-gating technique with 3D-translation compensation for free-breathing whole-heart coronary MRA. Magnetic Resonance in Medicine, 2009. 62(3): p. 731-738. 4. Piccini, D., et al., Spiral phyllotaxis: The natural way to construct a 3D radial trajectory in MRI. Magnetic Resonance in Medicine, 2011. 66(4): p. 1049-1056.

5. Krishnamoorthy, G., Smink, J., Kouwenhoven, M., Breeuwer, M, Anisotropic Field-of-Views in 3D Golden Angle Radial Imaging. Proc. Intl. Soc. Magn. Reson. Med. 2018, 2018. 4130.

6. Stehning, C., et al., Free-breathing whole-heart coronary MRA with 3D radial SSFP and self-navigated image reconstruction. Magnetic Resonance in Medicine, 2005. 54(2): p. 476-480.

7. Etienne, A., et al., “Soap‐Bubble” visualization and quantitative analysis of 3D coronary magnetic resonance angiograms. Magnetic Resonance in Medicine, 2002. 48(4 DOI - 10.1002/mrm.10253): p. 658-65866.

Figures

Fig 1. Schematic diagram of the proposed technique showing one of the heart-beats in an ECG-triggered acquisition. Three orthogonal projections (3D-PRONAV) are acquired before each interleaved 3D radial acquisition to estimate the 3D motion of the heart. Radial projections in the dummy cycle (5 nos.), 3D-PRONAV and the interleaved image acquisition share the same sequence parameters.

Fig 2. Example of the proposed motion estimation and coil selection from 3D-PRONAV. Plots of the Fourier transformed reference projections (black color) and the segmented blood pool (red color) of 3D-PRONAV in all anterior coil elements are shown in (a) followed by cross-correlation to estimate motion in mm (b). The accuracy of the estimated motion greatly depends on the accuracy of the segmented blood pool. Coil selection is done based on the maximum correlation coefficient (ρ) between the motion estimates and the COM computed in QBC (c). Selected coils are highlighted in (b) which had the largest ρ.

Fig 3. Comparison between SI motion estimated from 3D-PRONAV and the motion derived from diaphragmatic pencil beam navigator in one of the volunteers is shown in plot along with the regression analysis (red line) (a) the slope and the goodness-of-the-fit (R2) of the regression analysis for all the five volunteers is shown in (b). Note the slope varies greatly between subjects while the R2 is high and consistent indicating a good correlation between the SI motion estimated from 3D-PRONAV and the motion derived from the diaphragmatic navigator.

Fig 4. Example of 3D whole heart MRA data acquired in a healthy volunteer without motion correction (a), with rigid-body motion correction performed using the proposed technique only in SI direction (b) and rigid-body motion correction performed in all three orthogonal directions. The corresponding reformats computed using Soapbubble tool [6] to visualize major coronary branches are shown in (e to g). Note the improvement in visualization of arteries highlighted in red arrows compared to the images corrected only for SI motion.

Fig 5. Bar plot showing the length of the coronary branches measured using Soapbubble tool [6] in 5 volunteers. RCA = Right coronary artery, LAD = Left anterior descending artery, LCX = Left circumflex artery. RCA was visualized in 5/5 of volunteers and LAD/LCX was visualized in 4/5 volunteers.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2053