@inproceedings{oai:repo.qst.go.jp:00054064, author = {Kimura, Yuichi and Ogaki, Hiroto and Naganawa, Mika and Shidahara, Miho and Sakata, Muneyuki and Ishiwata, Kiichi and Suga, Mikio and 木村 裕一 and 大柿 宏人 and 長縄 美香 and 志田原 美保 and 坂田 宗之 and 石渡 喜一 and 菅 幹生}, book = {World Molecular Imaging Congress}, month = {Sep}, note = {The Logan graphical analysis (LGA) is a widely used algorithm for PET quantitative imaging. In LGA, a total distribution volume (VT) is derived as a slope in the linear relation between both radiopharmaceutical concentrations in tissues and arterial plasma after the time, t*, when the radiopharmaceutical distributes uniformly in regional tissues, and therefore t* should be specified automatically for every voxels. This study proposes a new algorithm for t* determination, ratLGA, based on the runs test, which is a nonparametric hypothesis test for randomness. In ratLGA, LGA was applied to compute residuals for the line estimation with a set of feasible t* for PET studies. Then, the runs test was employed to decide the randomness of signs in the residuals with a significance level of 0.05. ratLGA is expected to be robust against the noise observed in voxel-based tTACs. Using t* (5 to 80 [min]), ratLGA was applied to the simulated and clinical data for 11C-SA4503, which is a ligand for the sigma1 receptors (Sakata, NeuroImage, 2007). In the simulation, VT was widely ranged from 10 to 60 [mL/cm3] considering the kinetics of SA4503, and the linear regressions for estimated VT were y=0.94x-1.23 in ratLGA and y=0.81x+3.80 when t* was fixed at 30 [min] (Sakata, 2007). The underestimation was improved by ratLGA. ratLGA increased a contrast in clinical VT images as shown in figures. Figures A and B were VT in case of ratLGA and the fixed t*, respectively, and Figure C showed the difference. In conclusion, the runs test is promising to determine t* for LGA.}, title = {Determination of Starting Time in Logan Graphical Analysis Using the Runs Test for Quantitative PET Imaging}, volume = {2008}, year = {2008} }