Thoracic aortic atherosclerotic plaque is a major cause of ischemic stroke worldwide [1]. 3D black-blood vessel wall magnetic resonance imaging has been demonstrated to be a promising technique for the assessment of thoracic aortic plaques [2]. It usually uses a variable-flip-angle 3D turbo spin-echo (SPACE) sequence with a diaphragmatic navigator to minimize respiratory motion during free breathing. This technique faces two issues. The first issue is the low scan efficiency, resulting in a long scan time since only a small fraction of the acquired data is acceptable for reconstruction. The scan time is further prolonged when the imaging subjects have highly irregular breathing patterns or drift in respiratory motion occurs during navigator-gated scanning. The second issue is the robustness of the navigator method. This technique is based on the hypothesis that the motion between the diaphragm and heart has a strong correction. Then, heart motion can be corrected using the estimated superior–inferior (SI) motion of the diaphragm multiplied by an empirical correction factor (typically 0.6). However, the actual correction factor varies among individuals [3,4], which cannot be adjusted accurately. In addition, many studies have demonstrated that respiratory motion not only occurs in the SI direction [5] but also causes othersrigid motion or vascular deformation [6,7].

To address the above issues, the “self-gating (SG)” technique has been proposed to correct respiratory motion while maintaining a high acquisition efficiency. This approach extracts the respiratory motion displacement directly from the acquired k-space lines integrated during image acquisition and then corrects the motion in the acquired data using this information [8]. It allows a much larger acceptance window than conventional diaphragmatic navigator-gated acquisition with an increased scan efficiency close to 100%. In addition, SG lines are acquired in the heart region, allowing a more accurate estimation of motion displacement than diaphragmatic navigator-gated acquisition.

SG techniques can be divided into two groups. The first group employs radial trajectories, including 3D radial and stack of stars trajectories [9,10]. The k-space center points in the stack of stars or lines along the SI direction in the 3D radial direction are used as SG data. However, the radial trajectory requires precalibration due to hardware imperfections and has a very high computational complexity for reconstruction, particularly in iterative reconstruction of compressed sensing (CS) [11,12]. In the second group, the Cartesian trajectory is used in the SG technique by acquiring additional SG lines. It acquires data along spiral-like [13] or radial-like [14] interleaves on a Cartesian grid to achieve a variable-density sampling pattern, which can be used naturally for CS reconstruction. Most SG techniques use balanced steady-state free precession (bSSFP) or gradient echo (GRE) sequences, having bright blood signals and sufficient signal-to-noise ratios (SNRs) to extract motion displacement accurately. However, thoracic vessel wall imaging is a black-blood technique acquired with SPACE and may have a low SNR in the SG data relative to bSSFP or GRE sequences.

In this study, we propose a novel SG approach for thoracic vessel wall imaging with 100% scan efficiency. Imaging is performed using a modified SPACE sequence with the first two echoes collecting the SG data. The Cartesian k-space data are acquired using spiral profile ordering with a tiny golden-angle approach. The acquired data are binned into multiple respiratory states based on the motion displacements from the SG data. The images are reconstructed using L1 iterative self-consistent parallel imaging reconstruction (SPIRiT) [15] for each bin and are combined into one image after rigid image registration. Note that the use of a tiny golden angle is helpful to ensure that each bin has a reasonable calibration line number. The feasibility of the proposed method was demonstrated in 11 healthy subjects in about a 4 min free-breathing acquisition. The results were compared against conventional diaphragmatic navigator-gated acquisition with an average scan time of 8.5 min to assess the superiority of the proposed framework.