Carotid artery wall imaging requires high-resolution black-blood MRI, but its image quality is often degraded by neck motion. The neck area has very complex motion pattern due to superposition of breathing, abrupt swallowing and/or bulk neck shift. Previously, it has been demonstrated that it is feasible to correct both abrupt and bulk neck shift in carotid imaging using structured light^1,2. In this study, we aimed to separate different motion components and correct them for 2D carotid artery wall imaging, using non-marker-attached structured light system.

Motion detection: Using structured light, the height of surface (h) can be accurately calculated by the shift of the laser (d) projected on the subject, as h = dkH/ (D+dk)¹, where parameters k, H and D can be estimated by calibration using subjects with known heights. Structured light motion detection system was set up as Figure 1. To detect the neck motion during MR scans, a green laser cross was projected on the neck surface of the subject, which was captured by a MR-compatible camera placed in the bore of the MR scanner. The position of the laser was traced using pattern matching by LabVIEW (2013, USA) in real time. The coordinate of laser cross was recorded at every 2 ms. Network Time Protocol (NTP) was utilized to synchronize the time between the laptop recording the motion and the MR host. Also, the time delay caused by the camera and image processing was measured and compensated.

Image acquisition: A healthy volunteer was scanned for 2D black-blood carotid artery wall imaging using 8-channel carotid coil and 3T MR scanner (Philips, the Netherlands). 2D FFE sequence with inversion recovery was used: TR/TE = 100/5.4 ms, flip angle = 20°, FOV = 200 × 200 mm², resolution = 1.0 × 1.0 mm², slice thickness = 5 mm. For the first scan, the subject was instructed to avoid voluntary motion, while for the next ten scans, the subject was instructed to swallow during five scans and to raise head shortly during the other five scans.

Motion correction: For the ten scans with voluntary motion, the raw data was reconstructed using MATLAB (R2015a, USA) with the following three steps: (1) k-spaced lines acquired during abrupt motion was deleted and re-estimated using SPRiT algorithm³; (2) the initial bulk shift distance was estimated and every point in its vicinity was used as translation parameter to calculate reconstructed image sharpness; (3) bulk neck shift was corrected using optimized translation parameters with maximized sharpness of reconstructed image.

The respiration (16.9 ± 1.0 Rate/min), abrupt motion and bulk neck shift can all be separated from the optical motion detector (Figure 2). Averaged amplitudes for abrupt/ bulk/ respiration motions = 0.99/0.21/0.17 mm. For all the ten scans with voluntary abrupt motion, unconscious bulk neck shift were also detected after abrupt motion. The scan without voluntary motion (but have respiration motion) and four scans with abrupt motion affecting less than 7.0% peripheral k-space lines still had delineable contours of vessel wall. The other six scans had severe motion artifacts, which were successfully removed except for two scans with multiple swallowing affecting a large range of k-space lines (17.7% and 31.6%, respectively). One motion correction example was shown in Figure 3. With abrupt motion and bulk neck shift correction (shift distance optimized by maximizing sharpness), the motion artifact in vessel wall area was removed, enabling more accurate wall contour delineation. Carotid artery wall imaging has challenging motion issues due to the complex neck motion pattern. Our study demonstrated that non-marker-attached structured light system can effectively detect bulk neck shift, abrupt swallowing and respiration in the neck area. Respiration motion was not corrected since it introduced much less motion artifact than abrupt motion or bulk neck shift in carotid artery wall imaging. It was also shown that unconscious bulk neck shift often follows abrupt motion, both of which can be corrected by structured light system. The total duration of abrupt motion can affect not only the amount of motion artifact, but also the effectiveness of motion correction. Also, maximizing reconstructed image sharpness to optimize bulk shift distance can further improve the vessel wall delineation.