Free-breathing 3D high-resolution simultaneous grey-blood late gadolinium enhancement and MR angiography at 1.5T

Background: Cardiac magnetic resonance enables multi-contrast images, providing complementary information for the detection of various cardiac diseases. Coronary MR angiography (CMRA) offers detailed anatomical assessment of the coronary arteries and great vessels [1], while late gadolinium enhancement (LGE) remains the gold standard for non-invasive myocardial tissue characterization [2]. To optimize the image contrast of LGE, a grey-blood (GB) phase-sensitive inversion recovery (PSIR) sequence has been proposed by nulling the blood signal, which demonstrated improved results in endomyocardial scar depiction than conventional bright-blood (BB) imaging [3–5]. Here, we aim to achieve a one-stop 3D imaging of GB-PSIR and CMRA by proposing a simultaneous Grey-Blood and Bright-blOOd phase SensiTive inversion recovery (GB-BOOST) sequence at 1.5T.

Methods: The proposed GB-BOOST research sequence acquires two interleaved 3D volumes prepared with inversion recovery and T2 preparation (40ms) pulses respectively using bSSFP readout (Fig. 1). The inversion time is set to null blood signal. An image navigator enables 100% free-breathing scan efficiency [6]. A variable-density Cartesian trajectory with spiral-like profile order and golden-angle step is adopted with 4-fold undersampling [7]. Inline non-rigid motion corrected iterative SENSE with PSIR reconstruction is performed followed by offline high-dimensional patch-based low-rank denoising to finally produce T2prep GB-BOOST for vessel angiography and PSIR GB-BOOST for scar depiction [8,9].


All experiments were performed on a 1.5T MR scanner (MAGNETOM Sola, Siemens Healthcare, Forchheim, Germany). Clinical 2D BB- and GB-PSIR, and 3D GB-BOOST were sequentially scanned on 6 patients (3 males, 68±16 years) at approximately 10-20 mins after the injection of a gadolinium-based contrast agent (GADOVIST® 1.0 (gadobutrol), Bayer Healthcare) at a dose of 0.15 mmol/kg. A 3D CMRA was also performed on 3 patients for angiographic comparison. 3D sequences had a resolution of 1.2mm³.

Results: 3D GB-BOOST was successfully performed on all patients with an average scan time of 9.6±0.7mins. 3D PSIR GB-BOOST had image contrast comparable with 2D GB-PSIR, and 3D T2prep GB-BOOST showed good image quality comparable with 3D CMRA (Fig. 2). Images from a 74-year-old male patient are shown in Fig. 3. The 3D PSIR GB-BOOST and 2D GB-PSIR provided visually superior contrast than 2D BB-PSIR, presenting consistent findings regarding scar location. The left main coronary artery (LMS) occlusion was detected by both 3D T2prep GB-BOOST and 3D CMRA. The 3D PSIR GB-BOOST accurately depicts the scar area in the LMS territory.

Conclusion: The proposed 3D GB-BOOST sequence can achieve simultaneous whole heart GB-PSIR and CMRA in a single scan. 3D PSIR GB-BOOST achieves good contrast between scar, blood and myocardium, and 3D T2prep GB-BOOST is comparable with 3D CMRA for the depiction of the coronary arteries. These promising results warrant further studies in a larger patient cohort.

Figure 1. Sequence and reconstruction framework of the proposed GB-BOOST sequence with balanced steady-state free precession (bSSFP) acquisition. Two electrocardiogram-triggered interleaved 3D volumes are acquired with inversion recovery (IR) and T2 preparation (T2prep) respectively. Flip angle of 90° was used. The inversion time is set to null blood signal using simulation with blood T1. Inline non-rigid motion corrected iterative SENSE and PSIR reconstruction are performed followed by offline HD-PROST denoising on the T2prep and PSIR volumes to finally generate T2prep GB-BOOST and PSIR GB-BOOST images.
Figure 2. Representative images of a 66-year-old female patient acquired with different sequences. (A) LGE images acquired with 2D bright-blood PSIR, 2D grey-blood PSIR and 3D PSIR GB-BOOST in short-axis (SAX) and 2-chamber (2CH) views. The results showed negative LGE findings. 3D PSIR GB-BOOST had image contrast comparable with 2D grey-blood PSIR by nulling the signal of blood pool. (B) CMRA images acquired with 3D CMRA and 3D T2prep GB-BOOST with curved planar reconstruction to show left anterior descending artery (LAD) and right coronary artery (RCA) respectively. 3D T2prep GB-BOOST showed image contrast comparable with 3D CMRA, presenting good image quality for the depiction of the structure of both LAD and RCA.
Figure 3. Images of a 74-year-old male patient with known left-ventricle impairment vsecondary to coronary artery disease with chronic total occlusion of left main coronary artery. (A) Short-axis LGE images acquired with 2D bright-blood PSIR, 2D grey-blood PSIR and 3D PSIR GB-BOOST. There was a large area of subendocardial myocardial infarction in the left anterior descending artery (LAD) and left circumflex territory. Bright-blood PSIR presented a poor contrast especially for the detection of subendocardial and papillary muscle enhancement, while 3D PSIR GB-BOOST and 2D grey-blood PSIR had superior contrast, presenting consistent findings regarding the scar area across the heart. (B) Curved planner reconstructions of 3D GB-BOOST and 3D CMRA of the same patient. The occlusion of left main coronary artery was confirmed by the 3D T2prep GB-BOOST and 3D CMRA images (red arrows). In addition, PSIR GB-BOOST image with the same curved planner reconstruction showed scar area highly correlated with the left main coronary artery territory (red arrow).