A CNNs-based of Force and Torque Identification Model for Vascular Interventional Surgery Robot

Abstract

At present, vascular interventional surgery is still the most popular way to treat cardiovascular and cerebrovascular diseases, and it is currently the most recognized. During the traditional vascular interventional procedure, the surgeon needs to makes a judgment on the contact of the catheter in the human body with the inner wall of the vascular lumen according to the information of the catheter operation force felt by the operation, and then make an adjustment decision for the surgical operation. If the surgeon judges that the operating force is abnormal according to experience, he will stop pushing or twisting the catheter, and then perform the action of retreating to avoid damage to the vessel wall or rupture of the aneurysm. This paper proposes a variable-coefficient follow-up control method based on force and torque identification. By inputting the samples of forces and torques in the previous 50 sampling points into a convolutional neural network, the risk probability of the operating force and torque is obtained, which is used to calculate the following coefficient of master-slave system to ensure the safety of the surgery. Experiments have shown that the operating force and torque during surgery are reduced by 20.8% and 14.2%.