People who are interested in cardiac PET-MR. Both cardiac and respiratory motion may introduce bias in the estimation of the PET myocardial kinetic parameters. Simultaneous PET-MR makes it possible to perform MR-based PET motion correction. We have investigated the performance of MR-based motion correction on the estimation of myocardial PET kinetic parameters. Fig.1 shows the flowchart of PET-MR simulation, image reconstruction, and estimation of PET kinetic parameters. We performed simulation studies using an XCAT torso phantom [1], which includes heart, lungs, liver, and soft-tissue compartments and generates both cardiac and respiratory motion fields. The time activity curves of all the compartments were simulated according to a one-tissue compartmental model with realistic kinetic parameters and arterial input function from previously reported human ammonia perfusion studies [2]. The simulation data were equivalent to a 9-min dynamic PET scan with 0.5 mCi injection dose and the framing scheme of 8×5 sec, 4×10 sec, 2×20 sec, 1×40 sec, 1×2 min and 1×4 min. We simulated a defect in the left ventricle myocardium by lowering K₁ and k₂ values by 60% and 28%, respectively, of the original values. PET sinograms for each time frame were generated using a forward projection model, which incorporates attenuation, point spread function, and Poisson noise. Both scatter and random events were not accounted in the simulations. MRI simulation was performed using MRILAB [3] at 3T with published the T1, T2, and spin density values and standard GRE sequence. Three different types of motion were investigated: cardiac motion only, respiratory motion only, and both cardiac and respiratory motion. For each type of motion, a reference motion phase was selected. The motion fields transforming from a given motion phase to the reference motion phase were obtained by applying non-rigid Demon registration algorithm to the reconstructed MR images. For each time frame, PET image was reconstructed using Filtered Back Projection (FBP) for each motion phase. The resulting reconstructed PET images were then transformed to the reference motion phase using the motion fields. The motion corrected PET image for the reference motion phase in the time frame was then obtained by summing up all the transformed PET images. Finally, voxel-wise kinetic parameters were estimated using the motion corrected PET images by curve-fitting. For each type of motion, Fig.2 shows the reference (i.e. gating, only the PET events within the reference motion phase were used), motion-uncorrected, and motion- corrected static PET images. Fig.3 shows the estimated K₁ maps for the reference motion phase and each type of motion. Tab.1 shows the estimated kinetic parameters in different parts of the myocardium. Our results show that MR-based motion correction reduces motion blurring effects as well as the bias of the estimated kinetic parameters. Moreover, respiratory motion appears to contribute more to the blurring effect and the bias of the estimated kinetic parameters than cardiac motion if no motion correction is applied. We have shown that MR-based motion correction reduces the bias of estimated kinetic parameters in dynamic PET.