Background: The primary limitation of telesurgery is the communication latency. Accurate and detailed data are lacking to reveal the latency effects on surgical performance; furthermore, the maximum acceptable latency in telesurgery remains unclear.
Methods: Sixteen medical students performed an energy dissection exercise and a needle-driving exercise on the robotic simulator dV-Trainer(®), and latencies varying between 0 and 1,000 ms with a 100-ms interval were randomly and blindly presented. Task completion time, instrument motion, and errors were automatically recorded. The difficulty, security, precision, and fluidity of manipulation were self-scored by subjects between 0 and 4 (0 the best, 2 moderate, and 4 the worst).
Results: Task completion time, motion, and errors increased gradually as latency increased. An exponential regression was fit to the mean times and motions (R (2) > 0.98). Subjective scorings of the four items were similar. The mean scores were less than 1 at delays ≤200 ms, then increased from 1 to 2 at 300-700 ms, and finally approached 3 at delays above. In both exercises, latencies ≤300 ms were judged to be safe by all and 400-500 ms were accepted by 66-75 % of subjects. Less than 20 % of subjects accepted delays ≥800 ms.
Conclusions: The surgical performance deteriorates in an exponential way as the latency increases. The delay impact on instrument manipulation is mild at 0-200 ms, then increases from small to large at 300-700 ms, and finally becomes very large at 800-1,000 ms. Latencies ≤200 ms are ideal for telesurgery; 300 ms is also suitable; 400-500 ms may be acceptable but are already tiring; and 600-700 ms are difficult to deal with and only acceptable for low risk and simple procedures. Surgery is quite difficult at 800-1,000 ms, telementoring would be a better choice in this case.