Anatomical Feature Constrained Path Planning for Robot-Assisted Pedicle Screw Placement Surgery

Abstract In pedicle screw fixation, surgical robot and preoperative planning are enabling technologies to improve the accuracy and safety of pedicle screw placement. In this study, an automatic segmentation method for the pedicle and vertebral body is proposed based on the 3D anatomical features of vertebrae. Further, an optimal insertion path is obtained to balance the safety of pedicle screw placement and the vertebral-screw interface strength. The pedicle screw radius is then determined based on the pedicle radius. A classification method is proposed to assess the accuracy of path planning. Finally, the surgical robot's path can be updated based on the actual positions of the surgical robot and the patient. The CT data of 12 human vertebrae (T6−L5), 10 porcine vertebrae (L1−L5) and 5 ovine vertebrae (L1−L5) are used to validate the effectiveness of the proposed method. All pedicle screw placement paths are successfully generated, achieving an excellence or good rate of 98%. Ex vivo pedicle screw placement experiments are conducted on human spine phantom, porcine and ovine spines, and in vivo experiment is conducted on a Bama miniature pig. In the proposed method, both safety and accuracy of pedicle screw placement are improved. According to the widely recognized Gertzbein-Robbins classification, 93.18% of the outcomes achieve Grade A, showing promising potential in clinics.