Phase-specific CT alignment reduces respiratory errors in NH3 PET perfusion imaging

Objectives Respiratory motion leads to errors in cardiac PET/CT imaging. Errors are reduced by using a respiratory-averaged CT for attenuation correction but this increases CT-dose to the patient. It is hypothesized that alignment of a respiratory-gated PET study phase-by-phase with a single-phase CT can reduce respiratory-induced errors.

Methods 123 NH3 PET scans (70 rest) were gated at 15 breaths per minute ± 50%. Studies were acquired using our standard clinical protocol on a Discovery DRx PET/CT (GE Healthcare) in listmode with a respiratory trigger obtained via a RPM (Varian) monitor. A single fast CT acquisition (1.5s) was taken near full expiration immediately prior to (rest) or following (stress) the PET acquisition. Images were reconstructed with filtered backprojection and a 14mm Hann filter using a CT aligned with the ungated PET image. Respiratory phases were coregistered offline and the alignment transformation was used to measure motion. Studies with >8mm of displacement between two respiratory phases were reprocessed with the CT aligned separately to each phase. Images at expiration and inspiration were compared using 17-segment polar analysis and a normal database created from those studies with <6mm displacement.

Results Motion was greatest in the axial direction (4.85+/-1.76mm) and 8 scans (2 rest/6 stress) had >8mm displacement. Of these 8 scans, 5 had ≥ 2 connected segments with z-value differences between inspiration and expiration of >2 SD. Of these 5 scans, phase alignment resolved differences in 2 cases and reduced error in another. These results are similar to previous work with Rb82 cardiac PET

Conclusions Phase-specific alignment of CT reduces respiratory errors in NH3 cardiac PET with no increase in patient dose.