Zhang, Yitong;
(2025)
Workflow analysis for laparoscopic
surgery.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
Yitong_Thesis_final.pdf - Accepted Version Download (15MB) | Preview |
Abstract
This study investigated surgical workflow analysis by comparing frame-based and event-based methodologies. Initial research focused heavily on frame-wise classification and metrics. At the outset, we assessed traditional models against a proposed frame-based sequence-to-sequence model using both frame-based and event-based metrics across multiple datasets, including Cholec80 and an in-house Sacrocolpopexy dataset. Although conventional frame-based techniques generally performed well, they struggled with lengthy and complex surgical videos when evaluated on event-based metrics. Despite our proposed sequence-to-sequence model achieving superior event-based metrics, it still had limitations. Therefore, we introduced a new transition-retrieval configuration incorporating several innovative models: the TRN model for offline segmentation, and the enhanced ATRN model providing both offline/online segmentation and anticipation tasks. These methods showed improved performance in segmentation tasks by integrating transitions and minimizing frame-level noise. The study also underscored the importance of event-based metrics in capturing long-term temporal patterns and phase continuity, which is essential in the medical field.The Sacrocolpopexy dataset, which involves a unique type of surgery to current surgical workflow analysis benchmarks, contains surgeries of longer duration than those in existing benchmarks, thus increasing the likelihood of transition over-detection. We observed that the transition-retrieval configuration yields better results for this dataset. The research concluded that while traditional frame-based approaches are effective for quick evaluations, event-based metrics offer more detailed and accurate segmentation, which is essential for downstream surgical applications.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Workflow analysis for laparoscopic surgery |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Engineering Science > Dept of Computer Science |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10203752 |
Archive Staff Only
 |
View Item |
