A Force Feedback Pedical Screw Impantation Surgical Trainer with Automatic Vertebra Segmentation

Abstract

The spine is one of the most important parts of the human body. Spine diseases owning to aging or injuries would be treated by spine surgeries. However, the surgeries are highly dangerous and the surgical training can increase the success rate. In this paper, an interactive surgical simulation system for the training of pedicle screw implantation is introduced. We propose an automatic and efficient method to segment the vertebra from the CT images for building the vertebra model. The mean shift method which can cluster points having similar property into groups is used in data clustering. The clustered groups related to the bone region are chosen as the initial model. And then, we apply the Chan-Vese level set method which refines the vertebra contour by using the concepts of statistics. After segmentation, we then apply the marching cube algorithm to generate the 3D vertebra model, and pass it to a virtual surgical environment. A haptic device is adopted to interact with this model to fulfill the surgical simulation in the virtual environment. The average dice coefficient of the segmentation results is 0.92.