Motion Tracking In Virtual Surgery Adaptive Filtering Algorithm

Virtual surgery has extensive application prospect in pre-operative planning, training for operating skill of surgeon, surgical navigation, and design of artificial prosthesis. It combines the modern medicine with virtual reality, and multiple key techniques such as medicine, biomechanics, computer graphics, image processing, pattern recognition, and automatic control are integrated, constructing a virtual operating environment in computer where the surgery procedure is performed on a virtual patient by users. Human-computer interaction technique is one of the challenges in the development of virtual surgery systems, and an appropriate interaction scheme is crucial to the success of the systems. This paper is devoted to research on human-computer interaction technique used in a virtual cardiac intervention surgery system.Using motion tracking devices to track location and orientation of an object in virtual reality environment is a key interaction technique. FASTRAK is a 6DOF (6 degree of freedom) electromagnetic tracking system, which gathers the information of the location and orientation in three-dimension space with the electromagnetic induction theory. The wire is the main medical instrument to be simulated in virtual cardiac intervention surgery system, and using FASTRAK to simulate the wire and to perform location and interaction can meet the surgery simulation requirements, and help promote the simulation system. However, FASTRAK works on the theory of electromagnetic induction, therefore its requirement on operating environment is strict, and the precision and stability of the location data tend to be influenced by many external factors. In this paper, the major noise sources of FATRAK are studied in detail, and the noise model related to electromagnetic interference is described.Virtual cardiac intervention surgery demands real-time tracking and interacting with the wire accurately during the surgical operation, and functional modules like wire rambling and collision detection require highly on the precision and stability of the location data supplied by FASTRAK. Unfortunately, the intrinsic electromagnetic property of FASTRAK brings the location data noise interference, therefore real-time and effective motion tracking filtering is imperative. Traditional filtering methods can not suppress the noises in location data for they do not take the complicated environment into account. Adaptive Kalman filter is an effective dynamic filtering technology with high precision even when the noise statistics is unknown, but only Gaussian white noise is supposed to be included in the noise-signal model. To address the deficiency in the existing filtering methods, an adaptive Kalman filter algorithm based on wavelet analysis is proposed according to the actual demand of the virtual cardiac intervention surgery system. The discrete state equation and measuring equation for FASTRAK are established. To filter the color noise interfering the signal model, the measuring data are pre-processed by wavelet analysis, in which the color noise is suppressed with a threshold method. The pre-procession is followed by adaptive Kalman filter, in which the Gaussian white noise is filtered. The adaptive filtering algorithm is implemented in the software system. Simulation results show that the presented algorithm is effective and robust, improves the precision of FASTRAK data prominently.