Background: Detecting neural threats using electromyography (EMG) has gained recognition in the field of spinal surgery. To provide an efficient approach to detect neural threats during the operation of the spinal surgery robot, an automated method based the internal connection between EMG signal and neural proximity (NP) was explored by experiments.
Methods:A NP classifier was designed to distinguish the pattern of the threats.Then, it was evaluated in rabbit models in vivo. The experiments were conducted using 20 rabbits. In each rabbit, two puncture paths were created using a surgical robot. For each path, EMG signals were recorded at series of path-points with different neural proximities, and were constructed as datasets after data cleaning and processing. The proposed NP classifier was trained and tested on the datasets.Results: Classification accuracy of Path 1 and Path 2 were 99.1% and 94.0%, respectively.
Conclusion:This feasibility study proved that EMG can be used to detect the proximity of surgical instruments to nerve roots during robot-assisted spinal surgery. As the methods of detecting neural threats for surgical robots are still scarce, we believe this work will improve the clinical performance of spinal surgery robots and help the doctors to perform surgery safely.