Inter-frame motion correction for small animal PET imaging

Quantitative analysis of dynamic PET images provides invaluable information to biomedical research. Subject motion during the long dynamic image acquisition can cause motion artifacts that deteriorate quantitative analysis. Image registration technique can compensate for such motion and has been applied for human images especially in the context of dynamic contrast enhanced magnetic resonance imaging. The same principle can be applied to small animal PET imaging. In this study, we developed and evaluated several strategies using image registration based technique to compensate for inter-frame motion. Based on realistic simulation studies, it was determined that using principle component analysis to create a synthetic dynamic sequence as the reference for registration was the best performing motion compensation technique. Sub-millimeter accuracy was achieved even for the early five second frames. The motion correction technique was also able to improve the region of interest time activity curves and arterial input function (AIF). The proposed inter-frame technique generated promising results and warranted more extensive evaluation with animal PET data.