T1 weighted, black blood variable flip angle spin echo imaging utilizing variable density, distributed spirals and a low rank reconstruction
Kevin M Johnson1, Leonardo Rivera-Rivera1, and Patrick A Turski2

1Medical Physics, University of Wisconsin - Madison, Madison, WI, United States, 2Radiology, University of Wisconsin - Madison, Madison, WI, United States

Synopsis

Summary

Purpose

MRI presents an opportunity to explore vascular lesion interactions utilizing the combination of vessel wall imaging with measures such as 4D-flow and perfusion. To differentiate the pathologic vessel wall, images can be acquired after the administration of an exogenous contrast agent. Unfortunately, effective post-contrast black blood imaging is challenging due to the shortened T1 blood but also complex slow flow in disease such as aneurysms. Recent advances in black blood imaging include the introduction of variable flip angle fast spin echo imaging1 and black blood (BB) preparation modules using DANTE2; which can be combined3. Unfortunately, with the DANTE module tuned for maximum blood suppression the sequence is highly sensitive to patient and physiologic motion. Further, the use of variable flip angle echo results in significant blurring, especially in cases of altered T1. The purpose of this work is to investigate variable flip angle, distributed spiral4, fast spin echo; with the ultimate goal of relaxing demand on BB-preparation, achieving higher spatial resolution, and enabling the robust incorporation of ECG and T1 preparation.

Methods

Distributed spirals is a 3D Non-Cartesian trajectory which consists of a stack of spiral arms which are distributed onto with the kz phase encode set continuously and the angle between each arm on the kz being equal to the golden angle, 111.25°. It allows for acceleration in 3 directions and samples the center of k-space more frequently. One method of sampling k-space with this trajectory is to interleave kz phase encodes such that nearby kz phase encodes are sampled at roughly the same echo time, resulting in blurring along the z dimension. Alternatively, arms can be interleaved such that nearby phase encodes consists of a full set of echo times. This casts the echo train evolution as an undersampling artifact rather than a blurring artifact. Further, it means the center of k-space is sampled at multiple echo times. However, such data is well suited for reconstruction via low rank approximation along the echo train dimension5.

A variable flip angle, distributed spiral sequence was implemented on a 3T clinical scanner (MR750, GEHC, WI, USA). Gradients on x and y were rewound with crushers on the z axis. Outer volume suppression, and fast saturation proceeded the echo train. Images were first collected of a realistic neurovascular phantom (Shelley Medical, London, Canada) at net flow rates of 0, 250, 500, and 750 ml/min, roughly physiologic. Images were collected utilizing distributed spirals and Cartesian with matched parameters (1mm isotropic resolution, 24 echoes per excitation, TR=660ms, single channel coil). The echo spacing was slightly longer for spiral (6.6 vs 5.9ms). Black blood efficiency maps were computed by taking the ratio of the difference between flow and no-flow images over the no-flow images. Following analysis in phantoms, in-vivo feasibility images were collected in healthy volunteers with a 32ch head coil (Nova Medical, MA, USA). Distributed spiral images were collected with: (TR=550ms, ETL=24, 0.7mm isotropic resolution, scan time=5:15, 11,520 total arms, 2.5x under sampling) and compared to those achievable with Cartesian sequence of the same scan time (TR=550ms, ETL=24, 0.75x0.8x0.8mm3, 2x2 parallel imaging, scan time=5:15 min).

Results

Figure 1 shows maximum intensity projections of images collected in phantom studies with flow rates of and 0 and 250ml/min. For the flow rate of 250ml/min, Cartesian BB efficiency is heterogeneous and significantly correlated with the direction of flow, with limited flow suppression when the flow is in the A/P plane. At the same flow rate, the spiral images show less directional sensitivity and an overall improvement in suppression. This is quantified with a median BB suppression of 0.93 for spiral vs 0.69 for Cartesian. Figure 2, show the results of the low-rank reconstruction to fit echo train decay and phase evolution. Image contrast progresses from T1 dominated to T2 dominated as the echo train evolves. Utilizing the first echo from the reconstruction, high quality T1 weighted images of the whole brain black blood images can be acquired, as shown in Figure 3. Which as shown in Figure 4 appear to have better suppression of slow flow structures such as the transverse sinus compared to Cartesian imaging.

Discussion and Conclusions

In this work, we present a variable flip angle distributed spiral fast spin echo sequence. Instead of accepting blurring due to echo train magnetization evolution, this sequence allows the resolution of multiple echo times from a single scan. It uses rewound spiral trajectories in the x/y plane and crushers in z; which increases flow sensitivity. This is done without increasing the area of crushers which could lead to echo disturbing phase.

Acknowledgements

We gratefully acknowledge funding from NIH-NS066982 and GE Healthcare assistance and support.

References

1. Busse RF. Flow sensitivity of CPMG sequences with variable flip refocusing and implications for CSF signal uniformity in 3D-FSE imaging. Proceedings of the 14th Annual Meeting of ISMRM; Seattle, WA, USA. 2006. #2430

2. Li L, Miller KL, Jezzard P. DANTE-prepared pulse trains: a novel approach to motion-sensitized and motion-suppressed quantitative magnetic resonance imaging. Magn Reson Med 2012;68:1423–1438.

3. Xie, Y., Yang, Q., Xie, G., Pang, J., Fan, Z. and Li, D. (2015), Improved black-blood imaging using DANTE-SPACE for simultaneous carotid and intracranial vessel wall evaluation. Magn Reson Med. doi: 10.1002/mrm.257854.

4. Turley, D. C. and Pipe, J. G. (2013), Distributed spirals: A new class of three-dimensional k-space trajectories. Magn Reson Med, 70: 413–419. doi: 10.1002/mrm.24475

5. Tamir J, Weitian C, Lai P, Uecker M, Vasanawala SS, Lustig M. T2 Shuffling: Multicontrast 3D Fast Spin Echo Imaging. Proceedings of the 23rd Annual Meeting of ISMRM; Toronto, ON, Canada. 2015. #3399

Figures

Figure 1: Sagittal maximum intensity projection images of a realistic vascular phantom at flow rates of 0 and 250ml/min for distributed spiral and Cartesian based fast spin echo imaging with matched parameters. Flow suppression is directional and reduced in Cartesian sequence as compared to distributed spiral (arrows).

Figure 2: Resulting axial images reconstructed from the distributed spiral echo train utilizing a low rank constraint along the echo dimension. Echoes 1, 12, and 24 are shown. The evolution of contrast is clearly demonstrated and high resolution is maintained, despite the high level of undersampling.

Figure 3: Limited axial, sagittal, and coronal minimum intensity projections of distributed spiral sampling utilizing the first echo in the reconstructed series. Near complete flow suppression is demonstrated in both the arteries and veins.

Figure 4: Comparison of distributed spiral sampling utilizing the first echo in the reconstructed series to a sagittal acquired Cartesian sequence. Reformatted images are shown of the transverse sinus. Cartesian based sampling demonstrates residual signal which is not seen on the distributed spiral images.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4373