Research On Image Interaction And Constant Force Cutting Control Of Robot-assisted Spinal Laminar Decompression

There are some problems in the existing minimally invasive spinal surgery,such as relying heavily on the experience of doctors,long learning cycle,unable to quickly and large-scale promotion and rapid updating of the use of new surgical instruments and so on.Surgical robot has the advantages of good stability,high reliability,easy interaction,high precision and so on.It can assist doctors to realize high-precision intervertebral foramen shaping operation and improve the success rate of operation.The goal of this subject is to realize the technology of intervertebral foramen shaping assisted by minimally invasive surgery robot in the face of spine.This paper focuses on the automatic segmentation of information region of two-dimensional CT medical image,the planning of spinal cutting region under human-computer interaction technology and the compensation control of spinal physiological movement in dynamic environment.The purpose of this paper is to solve the problems of low efficiency of manual segmentation of medical images before operation,complex operation of existing electronic equipment and no aseptic operation and force position control under the influence of respiration in the hand in minimally invasive spinal surgery.Based on convolution neural network,a method for segmentation and reconstruction of spinal CT image is proposed.The preprocessing step of image enhancement is carried out to improve the recognizability of the original image.Li Chunming level set is used to construct the data set,which reduces the difficulty and learning period of manual labeling by doctors.On this basis,the derivative network of U-Net segmentation network is used to segment vertebral CT images.The target spine was reconstructed according to the results of segmentation.Based on the way of human-computer interaction,the method of cutting area planning is proposed.According to the specific use scene,the single and double finger features are improved,and the three-dimensional finger tip features are proposed.The distinguishing degree of gesture is improved,the feature is extracted by somatosensory sensor,and the hidden conditional random field and template matching are used for gesture classification.On this basis,the mapping relationship between gesture and planning operation is established.The method of constant force cutting of ultrasonic bone knife is put forward,and the scheme is designed by the combination of active and passive.The elastic clamping device and fuzzy stiffness control at the end of the robot are designed to improve the adaptability of the robot to the cutting environment of ultrasonic bone knife.This paper focuses on the improvement of the key problems in the traditional intervertebral foramen angioplasty based on the experience of doctors,combined with machine learning,human-computer interaction,active and passive control,etc.,focusing on the automatic segmentation of clinical traditional Chinese medicine images.Aseptic operation,dynamic environment compensation and other problems provide technical support for the application of robot-assisted intervertebral foramen angioplasty,which is of great practical significance.