Synopsis

Keywords: Data Acquisition, Blood

GRE-based vascular imaging suffers from the low acquisition efficiency. In this study, we presented a new 3D dual-contrast multishot-EPI based acquisition method called DESPAV for simultaneous MR angiography and venography. Full flow compensation for K-space center line and novel Center-out trajectory was implemented. A reconstruction pipeline was developed to correct for inter-shot phase errors, off-resonance induced artifact, and enable distortion-free multi-shot joint reconstruction. Preliminary results showed that DEPSAV can provide fast intracranial arterial and venous vasculature depiction with comparable contrast and image quality to 3D GRE-based TOF/SWI, while achieve ~3-fold reduction in acquisition time.

Introduction

Time-of-flight (TOF) MR angiography (MRA) and susceptibility-weighted MR venography (MRV) are widely used in the examination of cerebral vessels. However, the mainstream acquisition of MRA and MRV is based on gradient-echo (GRE) sequence and thus is limited in the acquisition efficiency, even with its multi-echo variant ¹. Recently, 3D multi-shot echo-planar imaging (ms-EPI) was introduced to acquire TOF or SWI images separately, with significantly reduced scanning time^2,3. Nevertheless, EPI has the following drawbacks for vascular imaging: (i) the flow-induced phase variation causes variable “ghost” artifact associated with the K-space trajectory; (ii) the long effective TE (TEeff) reduces the TOF contrast and introduces a blurring effect; (iii) low bandwidth in phase-encoding direction causes geometric distortion arising from B0-inhomogeneity; (iv) the shot-to-shot phase variation caused by eddy current and physiological noise needs to be resolved in ms-EPI. To overcome these problems and combined the acquisition of TOF and SWI images, we proposed the flow-compensated multi-shot Dual-contrast 3D EPI for Simultaneous MRAV (DEPSAV). The method consists of flow compensation, the center-out trajectory, and postprocessing workflow including phase correction and distortion-free reconstruction for highly-segmented ms-EPI. The feasibility of DEPSAV for fast cerebral angiography and venography was evaluated.

Method

Theory: Sequence design

• Flow compensation

First-order gradient moment nulling (GMN) was implemented in the flow-compensation block (FCB) to reduce the flow artifacts. The FCB was designed to fully flow-compensate at the time when the k-space center was filled in each shot (TE_eff). For each FCB, the fastest GMN-gradient was initialized with the general formula (Eq.1) and then determined by grid search, under the constraint of gradient hardware and peripheral nerve stimulation.

• Center-out reordering

To reduce the TE_eff and improve the SNR, the center-out sampling trajectory was implemented⁴. The TE_eff mainly depended on the duration of the prephaser (FCB-I), as shown in Fig.1(b). The k-space was split into two halves with inherently reversed phase-encoding traversal direction, which was essentially similar to the blip-up-and-down (BUDA) acquisition and enabled distortion correction with off-resonance estimation⁵.
The echo-train length (ETL) was limited in DEPSAV since the background suppression for TOF-contrast was impaired at long TR. In this study, the ETL was set to 9 with TR=49ms to compromise between the arterial and venous contrasts. With Ny=272 and centric segment overlapped, the total shot number was 16×2=32, and the effective undersampling factor R_eff≈30, which was a highly ill-posed inverse problem for reconstruction, let alone the capacity of parallel-imaging for further acceleration.
Theory: Reconstruction

• Shot combination

Considering FCB gradient moment is smaller and inter-shot interval (multiple of TR) is short compared with conventional ms-EPI, we assumed that: (i) the eddy current induced phase difference could be “self-navigated” corrected with the overlapped lines in k-space centeric segment as an extension to the point-by-point method by Hetzer (Eq.2)⁶. The “realigned” shots could be randomly shuffled or regularly combined as the input of the subsequent reconstruction to improve the conditioning of the problem and enabled robust solvability even with under-sampling (R_in-plane=2 in this study). (ii) The remanent 2D non-linear phase caused by physiological noise could be corrected by MUSSELS in the subsequent reconstruction pipeline⁷.

• Distortion-free reconstruction

The reconstruction workflow was similar to the series study of BUDA. Each half k-space was separately reconstructed as a partial-Fourier and under-sampling problem by S-LORAKS methods⁸.The B0-field map was estimated by TOPUP⁹. The data was transformed to extended hybrid-space for B0 correction. MUSSELS with “virtual-coil” framework for joint multi-shot reconstruction was utilized to eliminate the inter-shot phase error and improve the conditioning of the inverse problem.
Experiment
Simulation was performed to demonstrate the ghost artifact induced by off-resonance using center-out ordering. Standard resolution (0.8×0.8×1.2mm³) images of DEPSAV were obtained in phantom and in-vivo and compared with conventional GRE on a Siemens 3T Prisma system. The key parameters were shown in Table.1.

Results

Fig3.(a) showed that the ghost kernel of point-spread-function (PSF) caused by the highly discontinuous modulation with acquisition scheme of DEPSAV, which was different from the ‘split’ artifact of single-shot center-out EPI or the displacement of sequential EPI¹⁰. Fig.3(b),(c) showed the effect of ghost elimination and distortion correction by the field-map informed reconstruction.
Compared with GRE TOF-MRA, the TE₁ images of DEPSAV provided angiography with acceptable contrast between flow and tissue, although the background suppression was suboptimal due to the long TR.
Compared with GRE SWI, the TE₂ images of DESPAV achieved similar quality in depicting the cerebral veins. Moreover, the high-frequency k-space was filled later than TE_eff in the echo train of center-out ordering, which improved the capability of DEPSAV to capture smaller vessels and subtle structures.

Discussion & Conclusion

We presented DEPSAV, a dual-contrast multi-shot 3D EPI for fast and simultaneous acquisition of cerebral angiography and venography. Preliminary results show that the whole-brain TOF and SWI contrast images with 0.8×0.8×1.2mm³ resolution can be collected within 200s. The highly discontinuous k-space phase modulation due to the center-out reordering and multi-shot acquisition was successfully corrected by shot-shuffling and BUDA-style reconstruction.
Further improvements of DEPSAV such as multiple-overlapping thin-slabs techniques are still in progress. SVD-free optimization methods will improve the reconstruction speed. The higher isotropic-resolution acquisition will be evaluated on patients with cerebrovascular diseases to demonstrate its clinical values.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (82271985, 82001804, 8191101305), the Ministry of Science and Technology of China (2022ZD0211901, 2019YFA0707103), and the Natural Science Foundation of Beijing Municipality (7191003).