Image degradation caused by respiratory motion is an issue of concern in clinical positron emission tomography (PET) imaging. Compensation for respiratory motion requires an accurate motion sensor or a complicated data driven method. This thesis describes two new methods for respiratory motion gating and correction that relies upon information from the list mode data stream. The methods are based on a proposed hypothesis that geometric sensitivity varies along the z axis of a 3D PET scanner and can be used to compensate for respiratory motion. This hypothesis was developed into two directions: Geometric Sensitivity Gating (GSG) for respiratory motion gating and Geometric Sensitivity Correction (GSC) for respiratory motion correction. To test the proposed hypothesis, two steps have been undertaken in this thesis: simulation and clinical validation. Simulations were implemented by the use of Geant4 Application for Tomographic Emission (GATE) and NURBs-Based Cardiac Torso (NCAT) software packages to simulate the Phillips Allegro/Gemini PET Scanner and respiratory motion respectively. Patient data with both tumour and non-tumour cases were collected and clinical validation was performed on it. All data was collected in list-mode data format, which can be sorted into a sequence of frames. Then the proposed specific hypothesises GSG and GSC was applied to the sorted frames respectively. Simulation validations consist of geometric phantom and voxelised phantom validation. The geometric validation was implemented using a cylindrical phantom with a known sinusoidal oscillation. This oscillation was used as a representation of simple respiratory motion. When GSG and GSC were applied to the sorted list-mode data, they were shown to be able to gate and correct for the motion. For a more realistic simulation of respiratory motion, voxelised phantoms generated by NCAT were imported into GATE and employed to validate GSG and GSC, and gave encouraging indications for the application of respiratory motion gating and correction. Finally, when applied to the clinical patient data, it was demonstrated that the image degradation caused by respiratory motion were significantly reduced using GSG and GSC. Additional benefits of these two methods include: • No additional hardware device is required; • They only use list-mode data and are non-invasive; • There is no acquisition burden; • There is no additional patient preparation required; • There is no additional time required for clinical setup; This thesis demonstrates the proposed hypothesis of using the geometric sensitivity properties of a 3D PET scanner for respiratory motion compensation and proves this hypothesis using results from simulated and clinical studies.
展开▼
