CataractSAM-2: A Domain-Adapted Model for Anterior Segment Surgery Segmentation and Scalable Ground-Truth Annotation

TL;DR

CataractSAM-2 is a domain-adapted segmentation model for ophthalmic surgery videos that combines high accuracy, real-time performance, and an interactive annotation framework to facilitate scalable dataset creation and cross-procedural generalization.

Contribution

It introduces a domain-adapted extension of Segment Anything Model 2 with an interactive annotation tool, enabling scalable high-quality dataset creation and zero-shot generalization to related surgeries.

Findings

Achieves high-accuracy real-time segmentation in cataract surgery videos

Reduces annotation time through interactive prompts and mask propagation

Demonstrates strong zero-shot generalization to glaucoma procedures

Abstract

We present CataractSAM-2, a domain-adapted extension of Meta's Segment Anything Model 2, designed for real-time semantic segmentation of cataract ophthalmic surgery videos with high accuracy. Positioned at the intersection of computer vision and medical robotics, CataractSAM-2 enables precise intraoperative perception crucial for robotic-assisted and computer-guided surgical systems. Furthermore, to alleviate the burden of manual labeling, we introduce an interactive annotation framework that combines sparse prompts with video-based mask propagation. This tool significantly reduces annotation time and facilitates the scalable creation of high-quality ground-truth masks, accelerating dataset development for ocular anterior segment surgeries. We also demonstrate the model's strong zero-shot generalization to glaucoma trabeculectomy procedures, confirming its cross-procedural utility and potential for broader surgical applications. The trained model and annotation toolkit are released as open-source resources, establishing CataractSAM-2 as a foundation for expanding anterior ophthalmic surgical datasets and advancing real-time AI-driven solutions in medical robotics, as well as surgical video understanding.