Real-time Automated Performance Metrics for Virtual Temporal Bone Drilling
Abstract
Mastoidectomy is a difficult surgery to master and requires significant training. The primary goal of this work was to implement real-time performance metrics for an existing virtual surgical training simulator. To achieve this goal, image segmentation and bone- thickness estimation software modules were first interfaced to the system to provide patient-specific training data to the metrics system. Nine performance metrics were developed: burr selection, burr visibility, burr stroke consistency, burr orientation, excessive force application, dissected volume, drilling in proximity to critical structures, injury to critical structures, and critical structure exposure. Additionally, functionality was provided to view computed metrics, replay the virtual procedure, or re-attempt a portion of the procedure. Even with the metrics included, the simulator was able to maintain interactive performance. Construct validity of the metrics was established by demonstrating that they show a difference between junior and expert groups. Automated performance metrics may permit independent practice without requiring an instructing surgeon.