Research On Three-dimensional Augmented Reality Image Guided Technology For Abdominal Minimally Invasive Surgery

Minimally invasive surgery of the liver reduces trauma and shortens recovery time,but it also brings a series of problems.In the minimally invasive surgery,the doctor obtains images through the abdominal surgery image guided system.In the traditional surgery image guided system,the intraoperative image is provided by the laparoscope which has limited field of view and lack of depth information,and is separated from the preoperative planning.In order to solve these problems,this thesis applies augmented reality technology in image guided system and study augmented reality 3D image guided technology for minimally invasive abdominal surgery.This method breaks through the laparoscopic field of view limitation,establishes the visual connection of the preoperative intraoperative images,moreover,plans and displays the surgical path.First,this thesis study the method of semi-automatic segmentation of liver images in CT images.The drawbacks of the regional growth method were analyzed.Considering the actual situation of the liver image,the regional growth method judgment function was improved.This research established a liver image segmentation method,which combining the improved regional growth method,Snake algorithm and shape descriptor liver segmentation method,and compare this segmentation method with manual segmentation image.The advantages and disadvantages of the two-dimensional reconstruction methods of surface rendering and volume rendering are analyzed.The ray projection method in volume rendering is used to reconstruct the three-dimensional tissue models and analyze the reconstruction accuracy.Secondly,in order to achieve 3D-3D preoperative intraoperative image registration,the liver laparoscopic image was reconstructed in three dimensions.Reconstruction adopts the method of binocular visual feature registration.Firstly,the laparoscopic image is preprocessed.The filtering method combining smoothing filter and Laplace transform is designed to enhance the details of the image.For the occlusion of the image by the surgical instrument,The traditional inter-frame difference method is improved to remove the occlusion image.The Grabcut algorithm is applied to identify the liver region in the laparoscopic image.The Surf algorithm and FLANN matcher are used to identify and register the features.The feature pair is based on Lowe’s algorithm and Euclidean distance two-step screening.Based on the three-dimensional coordinates of the feature points of the polar geometry,the surface point cloud of the liver is reconstructed.Subsequently,the multimodal registration method of preoperative intraoperative images was studied,and this research adoptes the 3D-3D registration method.Firstly,the camera parameters are calibrated based on the checkerboard.The disadvantages of ICP point cloud registration are analyzed.Determining the two-step registration of the combination of contour registration and point cloud law.Three-dimensional registration of image guided system has been completed.Using the identification-based augmented reality display method,the transformation matrix is calculated and decomposed,and the AR display of the liver model is successfully realized.Finally,this research study the planning method of preoperative surgical path.According to the surgical path planning requirements,the path planning method of artificial potential field method was determined.The traditional artificial potential field method planning effect is analyzed,and the repulsive force field function is improved for its drawbacks,and the static establishment method of the traditional potential field is improved to dynamic establishment.A comparative analysis of the effects of the algorithm before and after the improvement,discusses the advantages and disadvantages of the improvement.