Abstract
Accurate measurement of the size of polyps during endoscopy is essential for optimising treatment strategies and assessing cancer risk. However, both patient outcomes and healthcare costs are influenced by the absence of a standardised method that provides both objectivity and precision. The primary aim of this study was to develop a quantitative method to measure polyp size and location both accurately and objectively, and to evaluate its feasibility. This method integrated an electromagnetic tracking sensor into a conventional endoscope. It then estimates the 3D coordinates of the polyp border based on the endoscope camera positions and the corresponding points of a polyp border in paired endoscopic images. The proposed method's performance is assessed in two settings: a simulated model and an ex vivo experiment.A simulated upper gastrointestinal endoscopy model serves as a foundation to firstly develop strategies for maximising accuracy with the proposed method, and secondly to evaluate the method's accuracy and reliability when actual values are available. This preliminary assessment yielded promising results concerning both reliability and accuracy in estimating polyp size and location.
This research evaluated a couple of approaches for identifying corresponding points in the context of the proposed method. The novel use of the shape-context approach, which relies solely on polyp segmentation, in an endoscopy scenario where features are not distinct, allows for the identification of corresponding points independently of the textureless nature of endoscopic images. The first ex vivo experiments, utilising various analyses, demonstrated the superior performance of the shape-context approach over the feature-based approach. In a subsequent ex vivo experiment, using multiple image pairs offered a potential means to compensate for polyp movements, suggesting a simple yet effective approach to address these issues. These results reaffirmed the reliability and validity of the proposed method in the ex vivo evaluation setting.
The results and validation experiments detailed in the thesis highlight the potential of the proposed method to improve upon the current visual estimation technique used in clinical practice. This experimental research represents a significant step forward in developing endoscopic devices with advanced quantitative features, serving as a proof of concept and opening doors to potential future clinical investigations.
Date of Award | 31 May 2024 |
---|---|
Original language | English |
Awarding Institution |
|
Supervisor | Raymond Lee (Supervisor), Simon Toh (Supervisor) & Martino Pani (Supervisor) |