Movement correction improves the quality of dynamic PET-CT images

Objectives Head movements occur often during PET-CT dynamic scanning and artifacts were induced in CT-based attenuation corrected PET images due to the mismatch between CT and PET images. These artifacts such as uneven tracer distribution cannot be eliminated by realigning PET images in different dynamic frames to a reference frame or coregistration to CT image alone. The aim of this study was to enhance the quality of dynamic PET images by correcting movements between CT and dynamic PET frames.

Methods A Hoffman Phantom filled with FDG was used for PET-CT scanning. Movements of translation, rotation and combination in different directions were made in different dynamic scanning frames to simulate head movements between frames. The CT image after stripping off the CT bed was first coregistered to different frames, then added back the attenuation of the CT bed, and used for reconstruction by applying the algorithm of attenuation weighted-ordered subsets expectation maximization (AW-OSEM), 6 iterations and 16 subsets. Realignments to the reference PET frame were done after reconstruction using coregistered CT image.

Results Artifacts due to movement in dynamic PET images were removed effectively after movement correction and reconstruction. ROIs were drawn on asymmetry artifacts for each frame to evaluate the correction of artifacts after CT movement correction and PET reconstruction. The sets of ROIs values were significantly different compared to reference frame before movement correction (p<5%) and became insignificantly after movement correction. Also the mean left-right asymmetry significantly decreased from 46.8% to 6.5%.

Conclusions Movement correction provides more reliable PET images in dynamic PET-CT scanning