Pixel-by-pixel parametric perfusion imaging can show the space distribution of blood flow rate and blood volume that is valuable for clinical diagnosis and treatment assessment. In this study, the feasibility of Nakagami imaging for suppressing attenuation effects was investigated. In order to avoid recirculation, a minimum-based method in determining curve interval for fitting was proposed and compared with threshold-based method. Three bolus kinetics models including gamma variate model, lognormal model and local density random walk model were used for curve fitting and compared. Parametric perfusion images were formed and discussed. Simulation and in vivo experiment were conducted for validation. The results show that Nakagami imaging is insensitive to attenuation effects. Minimum-based method is more efficient and robust in avoiding recirculation. Lognormal model provides optimal performance on curve fitting.