Surgical smoke generated during the minimally invasive surgery is normally regarded as an important aspect of surgical risk which could produce potential health problems for both clinicians and patients. Current researches in the area of surgical smoke simulation have limitations with regards to both accuracy and interactivity. In this paper, a novel approach based on the vortex particle method which combines the conventional fluid model and the meshfree method has been proposed. This method reconstructs the lost information during the simulation due to numerical dissipations by using a vorticity confinement force, and hence, successfully recovers the fluidic details including small-scale vorticities. GPU acceleration and image reconstruction based on medical images have also been employed to produce a patient-specific simulation. The results have shown to be high realistic and accurate, computational optimization and extensibility have also been achieved. Therefore, such simulation can be used in potential applications including surgical training and planning.