MICCAI 2021 Daily - Tuesday

large receptive field that can capture long-distance dependencies. We combine them together to form a two-step training. ” The team have seen promising results in an online surgery scenario. The processing of each frame takes less than one second, allowing a nurse or a robotic-assisted system, such as the da Vinci machine, to prepare the instrument ahead of time. “ I know a bit about the da Vinci machine, but I don’t have any experience using it, ” Kun points out. “ I’m a computer science student, so Professor Matthew Holden, who was a postdoc at Johns Hopkins, gave me the initial idea for this work and provides a lot of information. ” Another application for the work is in surgical coaching. When training a surgeon, it will be important to be clear about the phases of the surgery and the transitions between them, for example, how long they have to wrap things up before moving onto the next phase. Kun is keen for this work to rouse the interest of other researchers who may wish to turn their focus to the anticipation problem. “ The most exciting part of this work has been validation of the idea because before this paper we didn’t know if solving the anticipation problem was feasible using the deep learning method, ” he reveals. 11 DAILY MICCAI Tuesday Kun Yuan