Vision, gaze and advances in endoluminal control systems
File(s)
Author(s)
Sivananthan, Arun Ganan
Type
Thesis
Abstract
The field of luminal endoscopy has recently evolved rapidly, expanding both its diagnostic and therapeutic capabilities. As gastrointestinal cancer continues to rise in prevalence, endoscopy must advance to improve prevention, early detection, and minimally invasive treatment. Significant effort has focused on enhancing imaging technologies and therapeutic adjuncts, enabling earlier diagnosis and more effective management of luminal pathology. Despite these advances, the fundamental endoscope control system has remained largely unchanged for nearly half a century, relying on a bimanually operated flexible instrument controlled by dual wheels and cables.
The role of vision and gaze in endoscopy remains poorly defined yet represents an emerging area of research. Gaze analysis technology offers insight into visual search patterns during lesion detection. A systematic review demonstrates its potential as an objective tool to evaluate detection performance and assess optical technologies designed to enhance visualisation. A meta-analysis of real-time artificial intelligence–assisted systems confirms improved detection of precursor lesions. A cross-sectional survey of endoscopists further highlights persistent challenges in orientation and spatial control during procedures.
Review of current control systems and adjunctive tools reveals ongoing attempts to optimise endoscopic manipulation. However, the intrinsic flexibility of the endoscope presents mechanical challenges in delivering advanced therapy, contributing to a steep learning curve that confines complex procedures to a small cohort of skilled operators.
Recent work has explored robotic integration to enhance intuitive control. A systematic review of predominantly experimental robotic platforms identifies multiple systems at varying stages of development, though most remain costly and confined to trial settings.
This thesis introduces gaze control as a novel paradigm in endoscopic manipulation. Integration of eye-tracking technology with a robotised conventional endoscope enabled development of the first hands-free gaze-controlled system, demonstrating feasibility in benchtop and human trials. The refined iGAZE2 system represents proof of concept for a new approach to endoscopic control.
The role of vision and gaze in endoscopy remains poorly defined yet represents an emerging area of research. Gaze analysis technology offers insight into visual search patterns during lesion detection. A systematic review demonstrates its potential as an objective tool to evaluate detection performance and assess optical technologies designed to enhance visualisation. A meta-analysis of real-time artificial intelligence–assisted systems confirms improved detection of precursor lesions. A cross-sectional survey of endoscopists further highlights persistent challenges in orientation and spatial control during procedures.
Review of current control systems and adjunctive tools reveals ongoing attempts to optimise endoscopic manipulation. However, the intrinsic flexibility of the endoscope presents mechanical challenges in delivering advanced therapy, contributing to a steep learning curve that confines complex procedures to a small cohort of skilled operators.
Recent work has explored robotic integration to enhance intuitive control. A systematic review of predominantly experimental robotic platforms identifies multiple systems at varying stages of development, though most remain costly and confined to trial settings.
This thesis introduces gaze control as a novel paradigm in endoscopic manipulation. Integration of eye-tracking technology with a robotised conventional endoscope enabled development of the first hands-free gaze-controlled system, demonstrating feasibility in benchtop and human trials. The refined iGAZE2 system represents proof of concept for a new approach to endoscopic control.
Version
Open Access
Date Issued
2025-08-13
Date Awarded
2026-05-01
Copyright Statement
Attribution 4.0 International Licence (CC BY)
License URL
Advisor
Darzi, Ara
Patel, Nisha
Publisher Department
Department of Surgery & Cancer
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)