Study Of Navigation System For Percutaneous Coronary Intervention

The Chinese cardiovascular disease report shows that the prevalence and mortality of cardiovascular disease continue to rise year by year.Because of the advantages of small trauma,short recovery time and high safety,many doctors have used it as the main treatment for coronary heart disease.But traditional interventional surgery faces many challenges: a)During the operation,although the doctor is wearing heavy lead clothes,the head will still be exposed to X-rays,which will cause serious damage to the doctor's health;b)Due to the severe imbalance in the distribution of medical resources,most patients with cardiovascular disease cannot receive effective treatment.An effective way to solve these problems is to use robotic technology to assist in the interventional operation.An effective way to solve these problems is to use robotics to assist with interventional procedures.Robot assisted technology mainly includes interventional robot assisted operation and image-assisted visual assisted operation.In this paper,based on the urgent need of intraoperative image navigation system in clinical practice,and on the platform of vascular interventional surgery robot developed in the laboratory,combined with National Natural Science Foundation of China,"Study on the basic problems of minimally invasive interventional surgery robots"(61533016)and National Key R&D Program Digital Diagnostic Equipment R&D Special Project “Visually Guided Operation Master-Slave Motion Control and Software System Engineering”(2017YFC0110804),the methods of intraoperative interventional device tracking and motion path recognition are studied.In this paper,the denoising process of intraoperative DSA coronary images is first carried out,and then the related algorithms of interventional instruments and surgical path recognition are studied.The main research contents and contributions of this paper are as follows:Firstly,aiming at the noise problem of DSA images during coronary interventional surgery,this paper proposes an image denoising method based on codec structure,which uses the learning ability of convolutional networks to learn the mapping relationship between noise images and noise.Denoising DSA images to provide high quality video sequences for subsequent image navigation systems.Secondly,aiming at the low accuracy,real-time and robustness of traditional surgical methods for interventional surgical instruments,this paper proposes a new method of DSA image enhancement,which greatly improves the robustness of the detection network and an automatic real-time detection method based on RPN and resnet-101 was proposed,which significantly improved the real-time and accuracy of the surgical instrument detection network.Finally,the Canny edge segmentation method is proposed and a good result is obtained.Thirdly,aiming at the problem that the motion path of surgical instruments cannot be identified in the navigation system,this paper proposes a method based on full convolution for the recognition of surgical path.A deeper convolutional layer,a shortcut connection,a multi-image input,and a multi-feature fusion output are used,and a good path segmentation result is finally obtained.At last,the achievements are summarized and future work is presented.