Based On Motion Feature Extraction Of Pet Image Respiratory Motion Correction

PET (Positron Emission Tomography) is a kind of advanced technology for Nuclear medical imaging diagnosis. PET has some advantages with the functional imaging features compared with other structural imaging technology such as CT (computed tomography)and MRI (magnetic resonance imaging) in the early stage of cancer diagnosis, treatment plan and the evaluation of the treatment etc. However, PET has a disadvantage in the structural imaging, in addition, the artifact caused by respiratory motion will degrade the quality of PET image and impact the diagnosis of physician and the treatment of patient. Therefore, it is necessary to carry out the respiration motion correction.At first, the principle and features of PET imaging technology are introduced. The main factors that could degrade the PET image quality are discussed. Then the generation of respiration motion artefacts and the necessity of correction are illustrated.In order to acquire simulation data of the PET imaging for verifying the proposed methods, the structure of PET scanning devices and the working procedure are proposed. A PET simulation platform is composed of GATE(Geant4Simulation Application for Tomographic Emission),ROOT and STIR(Software for Tomographic Image Reconstruction) software packages. The effectiveness of the platform is verified through the simulation experiment.The feature of lung respiration motion is studied, and the respiration motion model is constructed. A kind of respiration motion correction method has been designed based on the motion character information abstraction. This method corrects the PET respiration image based on the data which have been extracted from the CT images. The abstraction of motion characteristics in this paper is based on the difference of CT images in different frame time. The decision of standard frame could generate the great effect to the correction consequence. Two different kinds of decision methods of standard frame are designed in this paper. The experiment proves that the dynamic standard frame method is more suitable for the correction of respiration motion. To better verify the effectiveness of correction method, RMCR (respiration motion correction rate) has been proposed in the paper. The proposed method of respiration motion correction based on respiration motion characteristic information abstraction has been implemented by Matlab software package. Some experiments are used to verify the method. Rigid and non-rigid motion data is used in the experiment.Rigid motion experiment is composed by three groups of experiments of geometrical phantoms, there are linear displacement, regular deformation and displacement deformation. NCAT respiration model simulation data has been used for non-rigid motion experiment. NCAT phantom which may reflect the non-rigid motion features of lung movement in respiration motion was generated through real CT data.