Skip to main content
. Author manuscript; available in PMC: 2022 Jan 1.
Published in final edited form as: Med Image Anal. 2020 Oct 17;67:101880. doi: 10.1016/j.media.2020.101880

Fig. 9.

Fig. 9.

Illustration of the temporal behavior before and after registration of a subject. Panels (a) and (b) illustrate coronal images for No-MoCo and after applying LiMo-MoCo, respectively. Panel (c) shows the signal intensity of a single voxel for No-MoCo, LiMo-MoCo and the LTI model. The position of the voxel is indicated with the red star in the coronal images and the dashed lines in panel (c) indicate the times at which the volumes in panels (a) and (b) were acquired. Before registration, the subject changes position between frames t2 and t4. During the motion event, the images become corrupted and all the signal is lost. LiMo-MoCo correctly aligns the sequence of volumes and inpaints the corrupted volume with the LTI model. The Total Variation metric after registration was 0.35 compared to 0.42 in the No-MoCo case.