首页|University Health Network Reports Findings in Artificial Intelligence (Developme nt, deployment and scaling of operating room-ready artificial intelligence for r eal-time surgical decision support)
University Health Network Reports Findings in Artificial Intelligence (Developme nt, deployment and scaling of operating room-ready artificial intelligence for r eal-time surgical decision support)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – New research on Artificial Intelligenc e is the subject of a report. According to news reporting originating in Toronto , Canada, by NewsRx journalists, research stated, “Deep learning for computer vi sion can be leveraged for interpreting surgical scenes and providing surgeons wi th real-time guidance to avoid complications. However, neither generalizability nor scalability of computer-vision-based surgical guidance systems have been dem onstrated, especially to geographic locations that lack hardware and infrastruct ure necessary for real-time inference.” The news reporters obtained a quote from the research from University Health Net work, “We propose a new equipment-agnostic framework for real-time use in operat ing suites. Using laparoscopic cholecystectomy and semantic segmentation models for predicting safe/dangerous (‘Go’/’No-Go’) zones of dissection as an example u se case, this study aimed to develop and test the performance of a novel data pi peline linked to a web-platform that enables real-time deployment from any edge device. To test this infrastructure and demonstrate its scalability and generali zability, lightweight U-Net and SegFormer models were trained on annotated frame s from a large and diverse multicenter dataset from 136 institutions, and then t ested on a separate prospectively collected dataset. A web-platform was created to enable real-time inference on any surgical video stream, and performance was tested on and optimized for a range of network speeds. The U-Net and SegFormer m odels respectively achieved mean Dice scores of 57% and 60% , precision 45 % and 53%, and recall 82% and 75% for predicting the Go zone, and mean Dice scores of 76% and 76%, precision 68% and 68%, and reca ll 92% and 92% for predicting the No-Go zone. After optimization of the client-server interaction over the network, we deliver a pre diction stream of at least 60 fps and with a maximum round-trip delay of 70 ms f or speeds above 8 Mbps.”
TorontoCanadaNorth and Central Ameri caArtificial IntelligenceEmerging TechnologiesHealth and MedicineMachine Learning
NETL
NSTL
