A team of Johns Hopkins and University of Maryland, Baltimore researchers has been awarded a $164,145 grant by the Maryland Innovation Initiative to commercialize Circlage, a cloud-based video analysis platform aimed at optimizing surgical care by harnessing artificial intelligence to assess surgeons’ skills and provide them with automated feedback.

Surgical outcomes can suffer when the expertise of the surgeons performing procedures is not fully developed or refined, sometimes leading to the need for further corrective treatments, added expenses, and an increased chance of medical malpractice claims. For example, early-career eye surgeons have a 9.3% greater frequency of high-risk complications than surgeons with at least 10 years of experience, according to a study published in Ophthalmology.

But by harnessing the power of AI, the Circlage team aims to consistently impart years of accumulated knowledge to less experienced surgeons without taking away valuable operating time from their more experienced mentors.

“Our collective vision is to eliminate unnecessary variation in surgical care resulting from differences across surgeons in their learning and skills,” says Shameema Sikder, the L. Douglas Lee and Barbara Levinson-Lee Associate Professor of Ophthalmology at the School of Medicine, the director of the Center for Excellence for Ophthalmic Surgical Education and Training at the Wilmer Eye Institute, an affiliate of the Malone Center for Engineering in Healthcare, and the leader of the Circlage development team.

To use Circlage, surgeons simply upload videos of their surgical procedures to receive AI-generated video skill assessments and rapid feedback. This new technology provides high-quality learning opportunities for trainees regardless of their location, delivering best-in-class expertise and individualized insights.

In the U.S. alone, surgeons already generate more than 400 million minutes of surgical videos per year—material that the research team has used to develop their AI model.

“We aim to build on the collective intelligence of experiential knowledge to make Circlage accessible and useful for all surgeons and patients across the world,” says Swaroop Vedula, an associate research professor at the Malone Center and a member of the Circlage development team.

With an initial focus on cataract surgeries, the team used neural networks to segment preexisting surgical videos into steps in a process called surgical phase recognition. They then used these segments, plus annotations from expert surgeons, to train deep learning algorithms to assess and rate newly uploaded videos’ accuracy concerning position and timing, quickly providing low-cost, efficient feedback to users in an easy-to-digest digital format.

Circlage’s current free-standing platform can also be used for post-surgery analysis and debriefing, supporting the continuing education of practicing surgeons by helping them to organize their own videos into personalized libraries for always-accessible self-analysis and skill review. The team also says that Circlage’s technology can be embedded into surgical microscopes for onsite use, automatically recording and analyzing all selected operations a surgeon performs.

Circlage is currently hosted on Microsoft Azure for residents at the Johns Hopkins Wilmer Eye Institute, but the researchers are working to embed and evaluate the platform in nationwide standardized eye surgery training programs, such as the one run by the Association for University Professors in Ophthalmology, the leading academic ophthalmology society in the country.

The team also plans to expand the platform to support procedures beyond eye surgeries, working with both academic and industry collaborators to identify key requirements for different end users, develop a prototype, and perform user testing.

“We intend to enable all surgeons globally to use Circlage for any—and every—procedure they perform,” says Vedula.

Vishal Patel, an associate professor of electrical computer engineering and a member of the Vision & Image Understanding Lab at the Whiting School of Engineering, is also involved in the project.

Learn more about the Maryland Innovation Initiative here.