Motion artifact is a major and inherent challenge in MRI. In brain imaging, the involuntary and voluntary head motion of patients can be approximated as rigid motion with 6 degrees of freedom (DOF) [[1], [2], [3]] in most imaging scenarios, which introduces phase shifts and rotations of the k-space during acquisition, and correspondingly results in artifacts. In neck or abdominal imaging, physiological motions, such as respiration, coughing, and swallowing, are the major causes of artifacts. These non-rigid physiological motions usually introduce complex k-space corruption and result in compound artifacts in reconstructed images.

To correct motion artifacts, several motion tracking methods have been proposed, including navigator-based, field detection-based, and optical motion tracking-based methods [1,2]. Among these methods, optical motion tracking methods have the unique advantages of high accuracy and precision, high tracking frame rate, and independence from the imaging process [1,4]. However, most existing optical tracking techniques are marker-based, including moiré phase [[5], [6], [7], [8], [9]], reflective spheres [10,11], and self-encoded markers [[12], [13], [14], [15]]. The marker-based optical tracking techniques require subjects to affix or wear markers during the entire scan. These markers may not only disturb the subjects but introduce bias when the marker fixation to the patient is imperfect [1,2], especially for uncooperative patients and children. Recently, a few markerless motion tracking systems, which measure the 3D facial surface based on the speckle-structured light, have been proposed and proven effective in brain imaging [[16], [17], [18]]. However, all these optical tracking devices were only demonstrated in rigid motion correction for brain imaging, while their capabilities in physiological motion compensation have not been investigated. Although some other techniques were proposed to extract respiratory motion by analyzing the illumination, 2D texture, or single-point depth changes of the subject's skin without need for measuring marker positions [[19], [20], [21]], their ability to track 3D rigid motion is limited in terms of accuracy and detection range.

In this study, we proposed a Structured Light Optical MOtion tracking (SLOMO) system based on parallel-line structured light, with only one camera and one projector, enabling it to detect both rigid and respiratory motions by sensing 3D-surface changes of subjects during MR acquisition. The SLOMO was validated for rigid motion correction in brain imaging and respiratory motion compensation in liver imaging.