Temporal image fractionation: rejection of motion artifacts in myocardial SPECT

We have developed a protocol, termed "temporal image fractionation," in which static myocardial perfusion SPECT studies are acquired as three-interval dynamic studies (three temporal frames, each consisting of a full projection set), utilizing continuous alternating detector rotation and a multi-detector camera. The frames are individually examined for motion by cine display, then summed together into a static SPECT file which is reconstructed with standard procedure. This approach offers three potential advantages in reducing or eliminating image artifacts resulting from patient or organ motion: (1) If severe motion occurs in one frame, only the remaining two are summed and reconstructed (motion-purging); (2) Alternating detector rotation reduces artifacts from mono-directional, drifting motion during acquisition (i.e., upward creep of the heart); and (3) Generally, with multiple rotations, motion is spread over a larger angular range and therefore has a lesser effect of the final reconstructed images.

These advantages are demonstrated and quantified in this paper using clinical data (A) and simulated motion on phantom data (B and C). In the phantom experiments, fractionated images were found to be 48.9%, 35.8% and 35.9% "more similar" to the original images than nonfractionated images for simulated 1.67-cm upward creep, 1.1-cm nonreturning axial motion and 1.65-cm lateral motion, respectively.

This protocol requires little extra processing and no final extra data storage compared to standard acquisition, and it has nearly eliminated instances in which a study had to be repeated due to patient motion. Step-and-shoot acquisition is not recommended in conjunction with this protocol, as it would lengthen the time necessary to obtain the same count statistics as in nonfractionated acquisition.