Correcting versus resolving respiratory motion in free-breathing whole-heart MRA: a comparison in patients with thoracic aortic disease

Abstract

Background: Whole-heart magnetic resonance angiography (MRA) requires sophisticated methods accounting for respiratory motion. Our purpose was to evaluate the image quality of compressed sensing-based respiratory motion-resolved three-dimensional (3D) whole-heart MRA compared with self-navigated motion-corrected whole-heart MRA in patients with known thoracic aorta dilation.

Methods: Twenty-five patients were prospectively enrolled in this ethically approved study. Whole-heart 1.5-T MRA was acquired using a prototype 3D radial steady-state free-precession free-breathing sequence. The same data were reconstructed with a one-dimensional motion-correction algorithm (1D-MCA) and an extradimensional golden-angle radial sparse parallel reconstruction (XD-GRASP). Subjective image quality was scored and objective image quality was quantified (signal intensity ratio, SIR; vessel sharpness). Wilcoxon, McNemar, and paired t tests were used.

Results: Subjective image quality was significantly higher using XD-GRASP compared to 1D-MCA (median 4.5, interquartile range 4.5-5.0 versus 4.0 [2.25-4.75]; p < 0.001), as well as signal homogeneity (3.0 [3.0-3.0] versus 2.0 [2.0-3.0]; p = 0.003), and image sharpness (3.0 [2.0-3.0] vs 2.0 [1.25-3.0]; p < 0.001). SIR with the 1D-MCA and XD-GRASP was 6.1 ± 3.9 versus 7.4 ± 2.5, respectively (p < 0.001); while signal homogeneity was 274.2 ± 265.0 versus 199.8 ± 67.2 (p = 0.129). XD-GRASP provided a higher vessel sharpness (45.3 ± 10.7 versus 40.6 ± 101, p = 0.025).

Conclusions: XD-GRASP-based motion-resolved reconstruction of free-breathing 3D whole-heart MRA datasets provides improved image contrast, sharpness, and signal homogeneity and seems to be a promising technique that overcomes some of the limitations of motion correction or respiratory navigator gating.

Keywords: Aorta; Dilatation; Image processing (computer–assisted); Magnetic resonance angiography; Motion.

Fig. 1

Representative images from a 66-year-old woman with ascending aorta dilation. Maximum intensity projection MRA images displayed as 3-mm thick slabs are shown in the candy cane view of the aorta reconstructed using motion-corrected (a) and motion-resolved (b) algorithms in end-expiratory phase. While the image quality of both datasets was rated the best, the improved sharpness and overall image quality achieved by the motion-resolved reconstruction can be clearly observed

Fig. 2

Representative motion-corrected (a, c) and motion-resolved (b, d) images from a 74-year-old woman (a, b) and a 59-year-old woman (c, d), both with ascending aorta dilation. Maximum intensity projection images displayed as 3-mm thick slabs are shown in the candy cane view of the aorta. Substantially improved image sharpness can be observed with motion-resolved reconstruction in both cases (b, d) along with improved signal uniformity especially in the first patient (a). Note that data from the same image acquisition are used but processed differently

Fig. 3

Left ventricular outflow track view of the heart visualizing the proximal segment of the right coronary artery. Motion-resolved reconstruction (b) provides improved sharpness (62.6%) compared to motion correction (50.9%, a)