Key Technologies Of Infrared Surgical Navigation Instrument

Along with social progress and the continuous development of medical technology, surgical operations increasingly focus on refinement and minimally invasive technologies. Based on the rapid development of computer technologies, precision manufacturing, medical imaging and image processing, computer-aided surgery becomes an inevitable trend of surgeries. Surgical navigation system is an application of computer-assisted surgery systems, which acts as the eyes of doctors, supplying real-time displays of the relative location and orientation between surgical instruments and the anatomical structures of lesion site for dortors. It can reduce surgical trauma, improve surgical precision and makes it possible for that conventional surgery can not achieve. It has become a hot and bright spot of medical industries. Because of national application and research needs of surgical navigation system, we develop a new generation of infrared surgical navigation prototype for clinical applications based on our previous research achievements.In order to comprehensively understand surgical navigation systems and grasp the research direction, this paper summarizes the features and advantages of surgical navigation system, showing the necessity for us to develop our own surgical navigation system. The major research contents and desired objectives of this paper are as follows:1) We studied the calibration algorithm of camera intrinsic parameters. First, without considering lens distortion, we use concentric circles as a calibration template. In addition to the advantages of planar template, the production of this template is much easier. The method is completely linear. It needs neither iterative optimization nor computation of the circular points, making the algorithm easier and robust. Then we calibrate radial distortions. In this paper, we adopt a curve fitting method. It uses a checkboard pattern that contains straight lines. After demonstrating that the image of a line under radial distortion is a circle, we use circle fitting algorithm to compute the circle parameters and combine with the relationship between the circle and radial distortions to finally calibrate the distortion parameters. This method reduces the complexity of iterative algorithms and avoids falling into local optimum. Thus, it is simple and effective. Furthermore, the method is more flexible because it separates the computations of lens distortions and the other camera intrinsic parameters.2) We studied the calibration algorithm for camera extrinsic parameters. Under the assumption that all the camera intrinsic parameters have been known, we focus on the calibration of camera extrinsic parameters, including the relative rotation matrix R and translation vector between the camera coordinate system and the world coordinate system. Hartely etc. proposed an optimization method, in which branch and bound method is used to search in rotation space and a series of second-order cone programming problems are solved to compute the optimal solution. In this paper, we improve Hartely's method, we use the plane coordinates instead of spherical coordinates to denote the image points, express the constraints as linear programming forms, and combine branch and bound method with linear programming to seek for global optimal solution, which has greatly enhanced the speed of the algorithm. Compared with Hartely's method, ours is about twice faster.3) We studied the algorithm of circle pose reconstruction. Since the positive surgical probe adopts infrared diodes as the markers which have circular shapes, the pose reconstruction of circle becomes very important. In this aspect, we propose a new pose reconstruction algorithm based on circular projection equation. This is a monocular reconstruction algorithm which has ambiguities in its solution. Although our algorithm has the same accuracy compared with the traditional methods, its computational complexity has been greatly reduced due to the matrix decomposition. Meanwhile, we give geometric explanations of the ambiguities of monocular circle reconstruction, and analysis the relationships and differences between sphere and circle, which is greatly superior to traditional algorithms in the theoretical completeness aspect.4) We developed the infrared surgical navigation system for clinical applications. We built a new generation infrared surgical navigation system based on our previous research. The hardware of this system consists of stereopsis locator, surgical instruments, and computer. The software of this system consists of image preprocessing, image feature extraction and correction, stereo vision computation, instrument segmentation, and real-time 3D visualization and navigation. Precision testing and simulated navigation experiments demonstrate that our system satisfies all the requirements of clinical navigation features. Furthermore, we carried out two clinical operations in Shanghai Ninth People's Hospital, and the results show that our system can basically meet the demands of clinical operations and its performance has reached the international standard.