Research On Key Technologies Of Liver Surgery Navigation System Based On Augmented Reality

During partial liver resection,the invisible internal anatomical structure of the liver directly affects the accuracy of the operation.At the same time,the surgeon needs to fit the patient’s preoperative CT or MRI and other two-dimensional image information with the patient’s intraoperative three-dimensional information,which requires higher surgical experience.The psychological burden of doctors during the operation is relatively large,and the information of intraoperative ultrasound images is difficult to understand,which limits the development of partial liver resection.The emergence of augmented reality technology provides a solution to this problem.This technology is to obtain a preoperative model after preoperative CT segmentation and reconstruction,and superimpose the preoperative model on the surgical scene,visually display the patient’s liver anatomy,and guide the doctor to complete the surgical task.Based on the analysis and explanation of the key technologies of augmented reality surgical navigation system,this paper conducts in-depth exploration and gives an improvement plan for two types of problems in the current navigation system,namely display mode and registration mode.In view of the traditional surgical navigation system adopting the display screen,the doctor needs to switch frequently between the patient and the display,which will cause unnecessary operation interruption.This paper improves on this by using an optical transmissive head-mounted display to directly project the patient’s preoperative image between the patient and the doctor’s eyes,which can significantly shorten the operation time.According to the different tracking systems,two tracking/registration systems are designed,namely the vision-based tracking system and the infrared-based tracking system.Aiming at the vision-based tracking system that realizes virtual and real model registration based on vision-related knowledge,the pinhole camera model and the logo recognition process are analyzed,and the logo is specially processed to improve the stability of logo recognition.In the infrared-based tracking system,the Holo Lens and virtual and real models are calibrated by hand-eye calibration and SVD decomposition.Finally,the tracking and registration process is realized.Liver tissue is soft tissue,and the accuracy is poor if pure rigid registration is used.This paper incorporates the biomechanical properties of liver tissue into the registration process,and proposes a non-rigid registration method.First,after analyzing the internal tissue of the liver,an assumption is made,and the liver tissue is regarded as a homogeneous and isotropic viscoelastic body.Then,to simulate the viscoelasticity of the liver,this paper uses smooth particle hydrodynamics to simulate the particle pressure,viscous elastic force and calculate the particle acceleration and then use the explicit solution method to update the particle position.Finally,in order to improve the search efficiency of the particle’s nearest neighbors,this paper constructs a KD tree data structure for accelerated processing.In order to complete the cutting operation during the operation,this paper proposes a multi-row progressive cutting surface drawing algorithm.The algorithm obtains realistic cuts by modifying the topological structure of the surface model.First of all,for the precise collision point between the surgical instrument and the model,this paper constructs a surface heuristic AABB collision detection algorithm,which can quickly obtain the collision primitives and obtain the accurate collision point.Subsequently,the cutting line is drawn using the moving vertex method.The process mainly includes four parts: searching for the nearest point,moving the nearest point,optimizing local primitives,and splitting the grid.The optimization of local primitives adopts the method of exchanging the vertices to make the primitives conform to the Delaunay triangulation criterion.Finally,after the cutting line is characterized by a Bezier curve,interpolation is performed between the tip of the scalpel and the model to complete the drawing of the cutting surface.In order to verify the contents of the above three chapters,this paper builds two surgical navigation systems and introduces the experimental process in detail.Subsequently,this paper divides the system error into registration error and model error from the perspective of error source,and designs experiments to evaluate each part of the error.Finally,using the isolated pig liver as the experimental object to complete the cutting experiment process assisted by the augmented reality surgery system,the results show that the system has a high guiding significance for clinical surgery.