AI video analysis enhances training in cataract surgery

Cataracts remain the leading cause of blindness globally, disproportionately affecting people in low- and middle-income countries. SICS is a widely used, cost-effective technique in these regions, but it lacks sufficient training infrastructure and quality assurance systems. 

While AI-based video analysis tools already exist for phacoemulsification — a more common method in high-income countries — no such tools had previously existed for SICS. That gap is now being addressed with the release of the “SICS-105” dataset, which includes surgical videos from 105 SICS procedures conducted at Sankara Eye Hospitals in India. 

The deep learning model MS-TCN++, developed by Prof. Jürgen Gall’s group at the University of Bonn, achieved over 85% accuracy in identifying different surgical phases, such as incision preparation and lens treatment. 

“The analysis of surgical phases is important because it enables a quantitative comparison between different surgeons, feedback on identified critical steps and the detection of deviations from surgical protocols. It is therefore the first step towards automatic assessment of surgical quality,” says Dr. Kaushik Murali, president of medical administration at the Sankara Eye Foundation India. 

Following the research team is working on developing algorithms to detect surgical instruments and intraoperative complications — further steps toward automated surgical quality monitoring. 

MEDICA-tradefair.com; Source: Universitätsklinikum Bonn