Researchers from the University of Washington and Rochester Institute of Technology have been awarded a patent for their groundbreaking technology, "Method for Classifying Images," an artificial intelligence-driven approach designed to transform how surgical videos are reviewed, shared, and utilized.
The patent represents a collaborative effort among investigators from the University of Washington Departments of Otolaryngology–Head & Neck Surgery, Electrical & Computer Engineering, and Neurological Surgery, as well as collaborators from Rochester Institute of Technology. The inventor team includes UW OTO Resident Lingga Adidharma, MD, Randall Bly, MD, Zixin Yang, Ph.D, Christopher Young, MD, Ph.D, Yangming Li, Ph.D, Blake Hannaford, Ph.D, Ian Humphreys, DO, FARS, Waleed Abuzeid, MD, Manuel Ferreira, MD, Ph.D, Kris Moe, MD, FACS, and Daniel King, MD.
Modern endoscopic surgical procedures often generate hours of video recordings. While these recordings contain valuable educational and clinical information, they are rarely reviewed or archived due to their massive file sizes and the significant time required to analyze them. As a result, much of the potential value contained within surgical video data remains untapped.
To address this challenge, the team developed a novel algorithm that automatically identifies and removes uninformative footage, such as blurry frames, camera repositioning, and lens-cleaning sequences. The technology then detects key surgical events, procedural steps, and instrument usage to create a concise 2–3-minute highlight reel that captures the most important aspects of a procedure.
The innovation has broad applications across surgical education, clinical practice, and patient care.
For surgical trainees, the technology offers a highly efficient way to review critical portions of an operation with faculty mentors. Similar to how athletes review game footage, residents and fellows can rapidly revisit high-yield surgical moments and receive targeted feedback without spending hours watching full-length recordings.
The technology also has the potential to improve clinical communication. Surgeons can use condensed procedural summaries to quickly convey operative findings and techniques to colleagues, supporting consultations, patient handoffs, and multidisciplinary collaboration.
In addition, summarized surgical videos may enhance patient and family education by providing a clear and accessible visual overview of what was accomplished during a procedure, helping to improve understanding and engagement in postoperative care.
The underlying research demonstrated that the algorithm preserved more than 84 percent of key surgical events while outperforming commercially available video summarization tools. These findings highlight the technology's ability to efficiently reduce video length without sacrificing clinically meaningful content.
This patent marks an important milestone in the application of artificial intelligence to surgical workflows and reflects the growing impact of interdisciplinary collaboration between surgeons, engineers, and computer scientists. By transforming hours of surgical footage into concise, meaningful summaries, this innovation has the potential to advance surgical education, improve communication among healthcare professionals, and enhance the patient experience.
Congratulations to the entire inventor team on this remarkable achievement and their continued contributions to innovation in medicine and technology.