Intraoperative control for robotic spinal surgical system with audio and torque sensing

In pedicle screw insertion surgeries, the most dangerous part is the screw path drilling process. In current surgeries, surgeons guarantee not to drill through the vertebra by their haptic and auditory sense and experience. In this paper, an intraoperative real-time control method for a Robotic Spinal Surgical System (RSSS) with state sensing is proposed. A drilling state recognition with Audio-Torque fusion is developed. The short-time average drilling torque and its amplitude are used to construct a reference torque, and classify the drilling states. Aim to audio signals, Support Vector Machine (SVM) is used to classify the patterns, and Mel-frequency cepstral coefficients (MFCC) is extracted to train the mode and predict testing samples. By setting a different priority level for each sensor, the fusion information is for precise intraoperative control in the screw path drilling.