A novel respiratory motion correction technique for myocardial perfusion SPECT

Objectives In myocardial perfusion SPECT, large motion of the heart during respiration can introduce artifacts into the images. Respiratory motion correction approaches usually simplify and correct the motion of the heart as rigid-body translations. However, for some patients, the motion of the heart may include significant amount of rotation, for example, rotation of the apex relative to the base. In this case, heart rotation correction can be critical.

Methods The challenges of using rigid-body registration for motion correction of the heart are that (a) the overall torso is not a single rigid-body, organs like the liver do not move in the same way as the heart and (b) the pivot of rotation of the heart is unknown. In this work, starting from the respiration gated images, we developed a rigid-body registration approach with two unique steps to address these challenges: (1) generate the summed images from the gated images, segment the heart and generate the region-of-interest (ROI) of the heart from the summed images and (2) estimate the pivot position for rotation correction of the heart. Using the XCAT phantom (Duke University), we simulated respiratory motion gated myocardial perfusion images with 10 bins per cycle. The simulated motion of the heart included cranio-caudal translation up to 2.0 cm and the rotation pivoting at the base up to 15 degrees.

Results The proposed approach effectively corrected the motion of the heart in the respiratory cycle. The residual motion after correction was minimal compared to the magnitude of the motion prior to the correction. The artifact introduced by large motion in the myocardial images was effectively eliminated.

Conclusions We developed a novel rigid-body respiratory motion correction technique that effectively corrected both translational and rotational motion of the heart. This technique is more desirable than the general approaches that only correct the translational motion of the heart