A generic respiratory motion model for motion correction in PET/CT

Respiratory motion causes different magnitude qualitative and quantitative artefacts in medical imaging and especially in emission tomography. Solutions presented to date include respiratory synchronized PET and CT acquisitions. In order to increase the signal to noise ratio of the synchronized images the use of non-rigid transformations during the reconstruction process accounting for the respiratory motion have been proposed. In the majority of this work the 4D CT images have been used for the derivation of the necessary deformation maps. The objective of this work is to use a global respiratory model to avoid unnecessary 4D CT acquisitions and associated patient dose as well as to improve the temporal resolution of the deformation matrices used in the PET image correction process. The global model is based on principal component analysis (PCA) and can be adapted to a given patient anatomy needing only two static CT images in combination with respiratory synchronised images of the patient surface. The global model was then used to generate n CT images corresponding to a 4D CT temporal resolution (n = 5, 10, 20, 30 and 50) and corresponding deformation matrices. These deformation matrices were then used to correct for the respiratory motion using an elastic transformation correction method during the reconstruction process. NCAT phantom data for two simulated patients and clinical data for six patients (four patients were used for the model creation and two patients for the method validation) were used to determine the optimal number of deformation matrices needed in the reconstruction process.