| Popis: |
Aiming at the research of precise control and human-machine collaboration in interventional surgery, a master-slave isomorphic coordinated control system of minimally invasive vascular interventional surgery robot is designed to meet the requirements of master-slave conformity and two-way information feedback. The robot prototype system can build precise positioning and control minimally invasive interventional surgery with analysis the kinematics of the guide wire to achieve precise positioning and precise control of the tip of the guide wire. After obtaining the clinically collected angiographic images of the vascular guide wire, the guide wire is segmented using MobileU-Net(MobileNet encoding and U-Net decoding) neural network structure. Compared with traditional pixel-based extraction method with median filtering, we get better results with less noise. The accuracy rate of our work is 1.75% much higher than only using U-Net in the guide wire segmentation. What's more, the speed of our network is 4ms much faster than U-Net on a recent GPU. The displacement path of the guide wire in the blood vessel is calculated, which shows that the average distance of the guide wire from the center line is less than 0.5mm. |
