Motion-corrected reconstruction of parametric images from dynamic PET data with the Synergistic Image Reconstruction Framework (SIRF)

Motion correction has been added to a PET-MR reconstruction tool, SIRF, by incorporating a registration package, NiftyReg. New functionality has been demonstrated in the context of estimating kinetic parameters in the left temporal lobe, comparing direct and indirect reconstructions and evaluating the impact of using motion correction.Principal component analysis was used to detect motion and to determine time frames, while STIR's parametric-OSEM was used to perform the motion-corrected direct parametric reconstruction.It was found that the variance in the left temporal lobe decreased when motion correction was performed, and the same was true of direct reconstructions compared to indirect.With SIRF, the entirety of the demonstrated functionality can be performed from a single Matlab or Python script.