OBJECTIVE: This study evaluated the impact of network latency on telestenting performance. BACKGROUND: The feasibility of long-distance robotic telestenting was recently demonstrated, yet the impact of network performance on telestenting remains unknown. METHODS: Ex vivo and in vivo telestenting models were constructed by connecting a robotic drive over a wired network to a robotic control system up to 103 miles away. During consecutive attempts to robotically wire a coronary artery, investigators randomly added signal latencies from 0 to 1,000 ms. Outcomes included wiring success, wiring time (time to advance wire to preselected target landmark), and perceived latency score (5 = imperceptible; 4 = noticeable but minor; 3 = noticeable; 2 = noticeable and major; 1 = unacceptable). RESULTS: Wiring success was achieved in 95 of 95 attempts in the ex vivo model and in 57 of 57 attempts in vivo. No significant difference in wiring time was observed across added latencies from 0 to 1,000 ms in the ex vivo (p = .64) or in vivo (p = .40) models. Compared to an added latency of 0 ms, perceived latency scores were not significantly different for added latencies of 150 and 250 ms (p = NS for both), but were significantly lower for latencies ≥400 ms (p < .001). CONCLUSIONS: Added latencies up to 250 ms were not associated with perceived latency, but latencies ≥400 ms were perceptible. Based on these findings, future telestenting studies should utilize networks with latencies ≤250 ms if perceived latency is to be avoided.
OBJECTIVE: This study evaluated the impact of network latency on telestenting performance. BACKGROUND: The feasibility of long-distance robotic telestenting was recently demonstrated, yet the impact of network performance on telestenting remains unknown. METHODS: Ex vivo and in vivo telestenting models were constructed by connecting a robotic drive over a wired network to a robotic control system up to 103 miles away. During consecutive attempts to robotically wire a coronary artery, investigators randomly added signal latencies from 0 to 1,000 ms. Outcomes included wiring success, wiring time (time to advance wire to preselected target landmark), and perceived latency score (5 = imperceptible; 4 = noticeable but minor; 3 = noticeable; 2 = noticeable and major; 1 = unacceptable). RESULTS: Wiring success was achieved in 95 of 95 attempts in the ex vivo model and in 57 of 57 attempts in vivo. No significant difference in wiring time was observed across added latencies from 0 to 1,000 ms in the ex vivo (p = .64) or in vivo (p = .40) models. Compared to an added latency of 0 ms, perceived latency scores were not significantly different for added latencies of 150 and 250 ms (p = NS for both), but were significantly lower for latencies ≥400 ms (p < .001). CONCLUSIONS: Added latencies up to 250 ms were not associated with perceived latency, but latencies ≥400 ms were perceptible. Based on these findings, future telestenting studies should utilize networks with latencies ≤250 ms if perceived latency is to be avoided.
Authors: Kazim H Narsinh; Ricardo Paez; Kerstin Mueller; M Travis Caton; Amanda Baker; Randall T Higashida; Van V Halbach; Christopher F Dowd; Matthew R Amans; Steven W Hetts; Alexander M Norbash; Daniel L Cooke Journal: Neuroradiol J Date: 2021-08-16