Automated classification of electrosurgical cautery state

Abstract

Introduction: In computer assisted surgery, it is sometimes necessary to detect when an activated electrosurgical tool comes into contact with a patient, known as the energy event. By continuously tracking the electrosurgical tools’ location using a navigation system, these energy events can help determine locations of sensor-classified tissues of interest. Our objective is to detect the energy event and settings of a cautery robustly and automatically with a technique that does not disrupt the surgical workflow. This study aims to demonstrate the feasibility of detecting the cautery state during surgical incisions. Methods: I detected changes in the current in the cautery cables using a current sensor and an oscilloscope. I implemented a custom 3D Slicer module that uses machine learning to automatically detect energy events, the cautery’s mode, and cautery settings with no change to the surgical workflow. Results: The model was robust in classifying each cautery state with high accuracy, regardless of the different tissue types and power level parameters altered by users during an operation. The model was also able to detect the power level of the cautery. Conclusion: The results demonstrate the feasibility to automatically identify when surgeons make incisions during their operation.