Research On Markerless Surgical Navigation Technology Based On Structured Light

With the application of surgical navigation system,surgery is getting more minimally invasive and accurate.Surgical navigation system tracks the location of patient and surgical instruments,it displays the relative position between surgical instruments and patient.At present,the mainstream optical surgery navigation track and locate with markers,which the doctor needs to click the markers in turn for registering.It need to register again when the markers drop.In order to overcome the deficiency of marker surgery navigation system,this paper designs a markerless surgical navigation system based on structured light,and propose the corresponding method of system calibration,3D data acquisition and surface registration.The main contributions of this paper are as follows:(1)Build markerless surgical navigation system based on structured light.The system combined with near infrared optical positioning and structured light 3D imaging.The near-infrared optical tracking system uses for surgical instrument positioning,and the structured light imaging system obtains 3D data of surface in patient.(2)Method of system calibration.System calibration is to determine the internal and external parameters of the binocular near-infrared camera and binocular monochrome camera,and the location of projection structured light sheet.The camera calibration board can calibrate the four cameras simply and effectively,which can be captured by both near-infrared camera and t monochrome camera.A structured light calibration board with white lines and background is designed to calibrate the structured light sheets,and it's easy to make.We calibrate the structured light with a calibrated binocular camera.The calibration process is simple.(3)Acquisition of 3D structured light data.According to the characteristics of the structured light stripe,we combine Steger's curve detector with zero-crossing detection method to extract the sub-pixel location of structured light stripe.We use gray and line-shifting coding method,and implement a stripe linking method to reduce the wrong coding which happens on the edge of coding region.(4)Automatic registration method of image space and real space.According to the characteristics of 3D data,we remove noise points which far away from the surface,and then down sample point cloud to reduce the data size to improve the registration speed.A rough registration method of point clouds based on affine transformation is proposed,and then the Interative closest point algorithm(ICP)is used to optimize.Finally,the navigation system realizes automatic markerless patient-to-image registration.Registration error is 3.331 mm and time consuming is 137 ms in the phantom experiments.It can make up for insufficient of marker navigation to meet the clinical needs.