Abstract

Background
Endoscopic procedures have evolved significantly with advances in imaging, robotics, and artificial intelligence. However, procedural complexity, anatomical variations, and the high cost of device testing remain substantial challenges. Digital twin technology, which creates dynamic virtual representations of physical systems, offers a promising paradigm for addressing these limitations in endoscopic practice.

Aim
This review examines current and prospective applications of digital twin technology in endoscopy, evaluates implementation challenges, and identifies future research directions that could transform endoscopic practice.

Method
A comprehensive narrative review was conducted, synthesising literature on digital twin applications in endoscopy across multiple domains, including gastroenterology, surgical endoscopy, and optical imaging. The review examined device-tissue interaction simulations, image enhancement systems, procedural planning tools, and real-time monitoring applications. Studies were identified through systematic searches of PubMed/MEDLINE, Embase, Scopus, Web of Science, and IEEE Xplore, conducted from inception to December 2025, using predefined search terms combining digital twin concepts with endoscopy-related terminology.

Results
Current applications include device-tissue interaction simulations for virtual testing, digital twin-guided image reconstruction in fibre endoscopy, and preliminary work on procedural planning and navigation assistance. Device-level twins and imaging chain corrections are emerging, though full-fledged patient-specific twins remain largely conceptual. Major barriers include data acquisition constraints, computational complexity, model validation requirements, and clinical integration challenges. Importantly, many current implementations more closely resemble digital shadows or digital models than true bidirectional digital twins, highlighting the need for a clear maturity taxonomy in this domain.

Conclusion
Digital twin methodology in endoscopy is in its early stages but demonstrates significant potential. Transitioning from proof-of-concept demonstrations to clinical adoption requires modular development, enhanced sensing capabilities, hybrid modelling approaches, and collaborative efforts among engineers, clinicians, regulatory agencies, and industry partners. Future development should embrace ante-hoc interpretability, generative AI, human-centric design principles, and metaverse-based collaborative environments to accelerate safe clinical translation.