Respiratory Motion Correction in Dynamic PET with a Single Attenuation Map

In addition to static tracer uptake values used routinely in clinical facilities, PET imaging can provide useful information on tracer kinetics via the use of dynamic acquisitions where a set of time frames are acquired starting from the injection/inhalation of the radiotracer. In lung studies, kinetic parameters, estimated from compartmental modelling, are however affected by respiratory motion. When only one attenuation image is available, most existing motion compensation strategies are not appropriate for the initial short time frames, especially as the activity distribution changes rapidly over the early part of the dynamic acquisition. This work presents a preliminary study to handle respiratory motion using a two-step process that uses gated dynamic data as input. We first use joint reconstruction of activity and motion on the entire gated PET data to estimate deformation fields. This allows the subsequent reconstruction of each time frame separately with motion compensation. We present results comparing on one hand the compartment model fit residuals with and without respiratory motion compensation and on the other hand the diaphragm position in non-attenuation corrected images and from this method.