부산대학교 도서관

With the advent of ultra-high-performance dedicated brain PET imaging systems, such as the NeuroEXPLORER (NX), which offer high spatial resolution, high sensitivity, and improved timing resolution, human motion has become a major limiting factor for brain PET quantification. This includes rigid and non-rigid motion in the head and neck regions. Several data-driven methods using PET emission data have been previously proposed and evaluated for motion tracking. However, the accuracy of these methods is susceptible to the early time frames of a dynamic imaging study. Benefiting from the high sensitivity, increased volume of lutetium-based crystal material, as well as improved TOF resolution of the NX scanner, this study investigates the feasibility of utilizing the lutetium background radiation for head motion tracking using experimental data. An image-based centroid-of-distribution (COD) approach and a sinogram-based principal component analysis (PCA) method were applied to the lutetium transmission data acquired with a human subject and their efficacy was validated using a United Imaging Healthcare (UIH) marker-less motion tracking (UMT) camera. The results suggested that both data-driven methods can detect moderate and large motion with a high confidence level, and the sinogram-based PCA approach showed superior sensitivity to the COD method for small-motion detection (