Qualitative and Quantitative Improvements for Positron Emission Tomography Using Different Motion Correction Methodologies

Positron Emission Tomography (PET) data suffers from low image quality and quantitative accuracy due to different kinds of motion of patients during imaging. Hardware-based motion correction is currently the standard; however, is limited by several constraints, the most important of which is retroactive data correction. Data-driven techniques to perform motion correction in this regard are active areas of research. The motivation behind this work lies in developing a complete data-driven approach to address both motion detection and correction. The work first presents an algorithm based on the positron emission particle tracking (PEPT) technique and makes use of time-of-flight (TOF) information from the scanner to enhance the performance of motion detection. The algorithm functions entirely in raw coincidence event data space and facilitates the detection of all kinds of motion, including global and local. The application of the algorithm was first studied on respiratory motion correction both in phantom and clinical studies. The performance was validated against the current clinical standard- external pressure sensor band, and the detected motion signals by the proposed algorithm correlated well with the band. The algorithm was also compared against two state-of-the-art algorithms, and it was found to outperform both. Next, the application of the algorithm was explored in head-neck motion detection and therefrom motion correction. This work adopted two different paths for motion correction, namely, conventional and deep learning approaches. The former was implemented by proposing a variation of the PEPT algorithm for raw PET coincidence event data registration. The latter was explored by means of training a deep convolutional neural network for PET image data registration. Both approaches were validated against current clinical standards and found to produce comparable results with increased computational efficiency. The motivation behind this work as a whole was to achieve qualitative and quantitative improvements of the PET data by implementing the proposed motion detection and correction approaches, as well as provide a detailed analytical study to support the ongoing research in this field.