Research On Key Technologies For Multimodal Image-guided Intervertebral Disc Ablation Navigation

Intervertebral disc ablation,as the least invasive surgical method for disc herniation,has become an important treatment method for clinical disc herniation.The surgical navigation system based on image guidance can provide the relative positional relationship between the surgical instruments and the patient during the operation,which can reduce the difficulty of intervertebral disc ablation surgery to a certain extent.However,the existing mainstream spinal surgery navigation systems are originally designed to guide pedicle screw placement and only use single-modal image guidance,which cannot be used for intervertebral disc ablation surgical navigation.Therefore,studying an intervertebral disc ablation navigation system based on multi-modal image guidance has important clinical significance.This article has done the following four tasks around the development of an intervertebral disc ablation navigation system:(1)Rigid-elastic hybrid registration of multi-modal spine images.A registration method for multimodal spine images based on joint estimation of an affine-elastic deformation fields is proposed.This method proposes a Multiple Affine Matrices Estimation(MAME)module to simultaneously estimate the affine transformation matrices of different vertebrae on the spine image;and proposes an Affine-Elastic Fusion(AEF)module for The fusion of the affine deformation field of the vertebrae and the elastic deformation field of the soft tissue around the vertebrae;a local rigidity constraint module(Local Rigidity Constraint,LRC)is proposed to constrain the deformation field of the vertebrae region and maintain the rigidity of the vertebrae;thus improving the accuracy of spine CT and Accuracy of MR image registration.This experiment used a high-resolution lumbar spine data set A,a low-resolution lumbar spine data set B and a cervical spine data set C.The registration results showed that the average Dice coefficients of the vertebrae were 0.9136,0.8160 and 0.8308 respectively;the average target registration error(Target Registration Error,TRE)are 2.05 mm,1.74 mm and 1.71mm respectively.(2)Registration of patient physical space and image space in spine surgery navigation system.A method of registering the patient’s physical space and image space based on a three-dimensional mobile C-arm X-ray machine is proposed.This method is based on the mechanical movement of a three-dimensional C-arm with an absolute position encoder and has high repeatability positioning accuracy.It tracks the C-arm through an optical positioning system and achieves the patient’s physical space through only two steps:preoperative C-arm calibration and intraoperative C-arm tracking.The automatic registration with the image space provides an accurate and universal automatic registration method between the patient’s physical space and the image space for the minimally invasive spine surgery navigation system.Experiments were conducted on an accuracy evaluation phantom using two different types of surgical instruments.The experimental results show that the overall registration errors of the two instruments are 1.32 mm and 1.62 mm respectively.(3)Dynamic tracking of patients in spinal surgery navigation systems.In order to perform non-invasive dynamic tracking of patients during surgery,a registration method based on the known three-dimensional mesh(3D Mesh)of the C-arm position and the 2D X-ray projection image was proposed.The test results show that the average absolute error is 2.3mm.(4)Development of a multimodal image-guided intervertebral disc ablation navigation system software platform.A label-based multi-thread sorting algorithm is proposed to improve the speed of image transmission and reading in the navigation system.Pseudo-window and hierarchical rendering technology are proposed to improve the performance of navigation system image display.An automatic registration method between the robotic arm and the image space was proposed,and the surgical path for the robotic arm to perform navigation planning was completed.The average maximum error of the entire system under the F2554-22 computer-aided navigation system positioning accuracy measurement standard proposed by the American Society for Testing and Materials is 0.88mm.