Supported by Monash University, Wenzhou Medical University Eye Hospital, Mohamed bi Zayed University of Artificial Intelligence, Airdoc and the Centre for Artificial Intelligence and Robotics of the Hong Kong Institute of Science & Innovation of the Chinese Academy of Sciences, the 5th APTOS Big Data Competition focuses on phase recognition of surgical videos using machine learning (ML). A subset of the OphNet dataset that comprises 496 cataract surgeries with detailed annotations of 35 surgical phases will be used.
Description
A key element of computer-assisted surgery systems is phase recognition of surgical videos, which identifies different steps and phases of a surgical operation. Existing phase recognition algorithms require frame-wise annotation of a large number of videos, which is both time- and money-consuming. What if we can use ML to automate phase recognition? We should be able to develop an effective tool for workflow optimization, surgical training and intraoperative assistance, enhancing patient safety and efficiency and ultimately paving the way for the advancement of tele-surgery, robotic surgery and AI-assisted surgery.
In this synchronous competition, you will build an ML model to automatically identify and analyze different phases in an ophthalmic surgery video. You will work with almost 500 cataract surgeries with detailed annotations of various surgical phases. Through accurate recognitions of various surgical phases, you will contribute to the standardization of ophthalmic surgery to improve surgical outcomes, optimize medical resource utilization and provide valuable training for both surgeons and humanoid robots in the future.
Timeline
- April 1, 2025 – Launch of Competition (Participants can click here to sign up for the Competition.)
- April 7, 2025 – Accepting Submissions
- May 31, 2025 – Entry Deadline & Team Merger Deadline
- June 15, 2025 – Submission Deadline
- June 22, 2025 – Submission of Codes (The top 5 finalists must submit their codes for review by 18:00 UTC+8.)
- June 28, 2025 – Prize Presentation
Cash Prizes
The winning team with the best scores on the private leaderboard are eligible to receive US$5,000 on the condition that the champion will share the tips and pearls in devising the ML algorithms for phase recognition in the Closing Session of APTOSxJSAIO 2025. S/he can either attend the conference in person in which his/her registration (worth at US$580) will be waived or s/he can choose to make a live presentation online at the scheduled time to interact with the audience. Participation in the Winner’s Share & Tell is a pre-requisite for receiving the cash prize.
Additional Opportunties
Participants with valid scores on the leaderboard can write to APTOS Secretariat to request for a free registration to attend the 10th Asia Pacific Tele-Ophthalmology Society Symposium in conjunction with the 6th Annual Meeting of Japanese Society of Artificial Intelligence in Ophthalmology that is worth at US$580.
Acknowledgment
We thank Monash University for labelling the datasets for the 5th APTOS Big Data Competition.
We are also grateful towards Alibaba Cloud and Tianchi, our sole sponsors that provide us with the technical support necessary for the Competition to take place.
