A generalizable foundation model for intraoperative understanding across surgical procedures

2026-02-14

AI Summary

This work addresses the challenge of limited generalization in existing AI models for minimally invasive surgery, which stems from variations in surgical procedures and practitioners across institutions. To overcome this, the authors propose ZEN, the first intraoperative video foundation model capable of unified representation across multiple surgical procedures. Trained on a diverse dataset of over four million frames spanning more than 21 procedure types, ZEN leverages a self-supervised multi-teacher distillation framework and introduces a standardized benchmark for downstream tasks. It consistently outperforms current methods across zero-shot, few-shot, frozen-backbone, and full fine-tuning settings, achieving state-of-the-art performance on 20 downstream tasks. These results demonstrate ZEN’s strong cross-procedure and cross-institutional generalization, establishing a new paradigm for intraoperative assistance and intelligent surgical training evaluation.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Computer Vision: Diffusion Models for Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

In minimally invasive surgery, clinical decisions depend on real-time visual interpretation, yet intraoperative perception varies substantially across surgeons and procedures. This variability limits consistent assessment, training, and the development of reliable artificial intelligence systems, as most surgical AI models are designed for narrowly defined tasks and do not generalize across procedures or institutions. Here we introduce ZEN, a generalizable foundation model for intraoperative surgical video understanding trained on more than 4 million frames from over 21 procedures using a self-supervised multi-teacher distillation framework. We curated a large and diverse dataset and systematically evaluated multiple representation learning strategies within a unified benchmark. Across 20 downstream tasks and full fine-tuning, frozen-backbone, few-shot and zero-shot settings, ZEN consistently outperforms existing surgical foundation models and demonstrates robust cross-procedure generalization. These results suggest a step toward unified representations for surgical scene understanding and support future applications in intraoperative assistance and surgical training assessment.

Problem

• intraoperative understanding
• surgical generalization
• foundation model
• visual interpretation variability
• surgical AI

Innovation

• foundation model
• surgical video understanding
• self-supervised learning
• multi-teacher distillation
• cross-procedure generalization