At the 2024 Society for Nuclear Medicine and Molecular Imaging (SNMMI) meeting held in Toronto, Canada, from June 8-11, Caitlin Randell, a PhD student in Biomedical Engineering, and Alan McMillan, PhD, presented their research on enhancing cardiac PET imaging. This study addresses the significant challenge of capturing high-motion patient anatomy with the inherently low temporal resolution of PET imaging, crucial for diagnosing and treating heart disease, a leading cause of death worldwide.
The research introduces a novel, contactless, fully-automated data-driven dual gating algorithm for cardiac and respiratory motion correction in PET images. Using clinical [13N]-ammonia myocardial perfusion rest and stress scan datasets obtained under an IRB-approved protocol, the team developed an approach that utilizes ultrashort framerate images generated through a fast offline listmode reconstruction. By employing principal component analysis and frequency analysis, the team derived motion traces that allowed for the classification of each time point into discrete cardiac and respiratory states. This method enabled the reconstruction of PET images that captured both cardiac and respiratory motions, providing a comprehensive view of heart dynamics throughout the cardiac cycle without requiring any external signals such as EKG.
The results of this study demonstrated the effectiveness of the algorithm-derived gating signals compared to conventional EKG gating. Statistical analysis showed no significant differences between the heart rates calculated by the algorithm and those obtained via EKG. Additionally, the contactless dual-gated images exhibited improved contrast and image quality over non-gated and single-gated images, especially in cases of poor-quality EKG data. The findings suggest that this automated approach not only matches the performance of traditional EKG gating but also offers additional benefits by incorporating respiratory gating, thus enhancing the robustness of cardiac PET imaging.