Objective assessment of robotic surgical skill using instrument contact vibrations.

BACKGROUND: Surgical skill evaluation ordinarily requires tedious video review and survey completion, while new automatic approaches focus on evaluating the quality of the surgeon's movements in free space. Robotic surgical instrument vibrations are simple to measure and physically correspond to how roughly instruments are handled, but they have yet to be studied as a measure of technical surgical skill. METHODS: Thirteen surgeons used a robotic surgery system (da Vinci S by Intuitive Surgical) to perform four trials each of peg transfer (PT), needle pass (NP), and intracorporeal suturing (IS). Completion time, instrument vibrations, and applied forces were measured for each trial; root mean square (RMS) and total sum of squares (TSS) were calculated from both the vibration and force recordings. Four experienced surgeons blindly assessed the task videos using a Global Rating Scale (GRS), and skill metrics were compared between the eight novices and five experienced participants. Stepwise regression was performed to predict GRS score from objective skill metrics. The concurrent validity of each metric was evaluated using receiver operating characteristic (ROC) analysis. RESULTS: The GRS demonstrated excellent internal consistency (Cronbach's α = 0.91) and strong inter-rater reliability (ICC = 0.84). Compared to novices, experienced surgeons earned higher GRS scores and performed tasks with lower vibration magnitudes, lower forces, and shorter completion times in 15 of 18 task-metric combinations (p values ranging from 0.042 to <0.001). ROC analysis demonstrated that including vibration and force magnitudes along with completion time in skill prediction models improves the objective classification of subjects as novice or experienced for all tasks studied (PT: 90% sensitivity, 75% specificity; NP: 85% sensitivity, 84% specificity; suturing: 100% sensitivity, 100% specificity). CONCLUSIONS: RMS and TSS instrument vibrations are novel construct-valid measures of robotic surgical skill that enable the development of objective skill assessment models comparable to observer-based ratings.

A practical system for recording instrument interactions during live robotic surgery.

We have developed a system for measuring and recording the high-frequency vibrations that characterize instrument interactions during minimally invasive robotic surgery. Consisting of simple circuitry and a DVD recorder, this system is low-cost and easily implementable, requires no sterilization, and enables measurement of a validated, objective technical skill metric in both the simulated setting and the operating room. The vibration recordings of fourteen sleeve gastrectomies were processed by segmenting the operation into seven phases and calculating the root mean square (RMS) vibration within each phase. Statistical analysis showed that the observed differences match expectations drawn from knowledge of the operation, substantiating the premise that RMS vibration provides a good measure of the intensity of instrument interactions during live robotic surgery.

In vivo validation of a system for haptic feedback of tool vibrations in robotic surgery.

BACKGROUND: Robotic minimally invasive surgery (RMIS) lacks the haptic (kinesthetic and tactile) cues that surgeons are accustomed to receiving in open and laparoscopic surgery. We previously introduced a method for adding tactile and audio feedback of tool vibrations to RMIS systems, creating sensations similar to what one feels and hears when using a laparoscopic tool. Our prior work showed that surgeons performing box-trainer tasks significantly preferred having this feedback and believed that it helped them concentrate on the task, but we did not know how well our approach would work in a clinically relevant setting. This study constituted the first in vivo test of our system. METHODS: Accelerometers that measure tool vibrations were mounted to the patient-side manipulators of a da Vinci S surgical system. The measured vibrations were recorded and presented to the surgeon through vibrotactile and audio channels while two transperitoneal nephrectomies and two mid-ureteral dissections with uretero-ureterostomy were completed on a porcine model. We examined 30 minutes of resulting video to identify and tag manipulation events, aiming to determine whether our system can measure significant and meaningful tool vibrations during in vivo procedures. RESULTS: A total of 1,404 manipulation events were identified. Analysis of each event's accelerations indicated that 82 % of these events resulted in significant vibrations. The magnitude of the accelerations measured for different manipulation events varied widely, with hard contact causing the largest cues. CONCLUSIONS: This study demonstrates the feasibility of providing tool vibration feedback during in vivo RMIS. Significant tool vibrations were reliably measured for the majority of events during standard urological procedures on a porcine model, while real-time, naturalistic tactile and audio tool vibration feedback was provided to the surgeon. The feedback system's modules were easily implemented outside the sterile field of the da Vinci S and did not interfere with the surgical procedure.