Data-Driven Patient Motion Compensation in Cardiac Positron Emission Tomography

Positron emission tomography (PET) is a molecular imaging modality that has been demonstrated to be a powerful, non-invasive, tool for the assessment and diagnosis of cardiac pathologies like coronary artery disease. The accuracy of these clinical examinations for detecting and prognosticating disease can be marred in cases where patient motion is severe. Clinical use of motion tracking/compensation tools, however, is relatively uncommon partly due to the increases in complexity and time of patient setup prior to imaging. The purpose of the work described here was to develop and evaluate new methods of patient motion detection and compensation in the context of cardiac PET imaging studies that are less complex than standard commercial options in the hope of reducing barriers to clinical adoption. The proposed methods are based on measuring and tracking the motion of a low-activity radioactive marker placed on patients using the positron emission tracking (PeTrack) algorithm. Motion information was employed to compensate and/or correct for either respiratory or whole-body patient motion. The performance of PeTrack for respiratory tracking and motion compensation was evaluated in a clinical population in comparison with a commonly used commercial optical tracking device. Within a practical comparison framework PeTrack was shown to perform comparably to the commercial system. From this comparison shortcomings of both PeTrack and the commercial system were identified; knowledge of the former can inform future development and improvement. A method for whole-body patient motion correction (WBMC) in static cardiac perfusion studies using PeTrack was developed. Motion corrected images demonstrated significantly less blurring of the myocardial walls and improved contrast. Relative perfusion measurements among the clinical data sets were not significantly affected although the extent of patient motion was limited. The WBMC algorithm was extended for dynamic acquisitions used for quantification of myocardial blood flow. Motion detection and estimation with PeTrack was compared to that of another data-driven motion tracking algorithm within a clinical population. Body motion estimation with PeTrack was more robust than the alternative method. Motion correction using PeTrack demonstrated improvement among various quality indicators of the kinetic modelling used to estimate blood flow.