Data driven respiratory gating applied to whole body, 3D clinical FDG PET acquisitions

Objectives Respiratory gating divides a PET acquisition into a series of near motion free frames. Typically, hardware is used to measure the respiratory state of the patient during the PET acquisition. Data-driven gating methods derive a respiratory signal from the acquired data itself. In this work we have extended a data-driven respiratory gating method and applied it to 3D, whole body clinical FDG PET acquisitions.

Methods The gating method, previously demonstrated with 2D, single bed-position data, was modified to accept restricted lines of response by reducing the axial acceptance angle. Lung bed positions were processed first, and the resulting respiratory frequency estimates were used to gate the remaining bed positions. As the data-driven signal does not identify the direction of motion, a registration based method was developed to align the direction of motion across all bed positions. Whole body PET acquisitions were obtained for 11 clinical patients, and hardware based respiratory signals were also collected for comparison. All data were retrospectively gated offline using both hardware and data-driven methods, and gated 3D sinograms were restored to the PET camera for reconstruction. Total displacement was determined in the gated images by measuring the change in centre of mass within 4 regions of interest in each patient.

Results Good agreement was found between the hardware based and data driven respiratory curves for bed positions over the abdomen and lower lungs, with on average, less than 23% difference between the signals. The mean displacement for the data driven gated images was 10.3mm, and 9.1mm for the hardware gated images. No significant difference was found between the two gating methods when comparing the displacement values.

Conclusions The extended data driven method provides a feasible alternative to respiratory gating whole body, 3D clinical PET acquisitions