University of Twente Student Theses
As of Friday, 8 August 2025, the current Student Theses repository is no longer available for thesis uploads. A new Student Theses repository will be available starting Friday, 15 August 2025.
Revealing the margins : Towards ex vivo image-guided surgery in oral cancer resection : Facilitating intraoperative assessment of resection margins during oral squamous cell carcinoma surgery using automatic segmentation and 3D visualisation on ex vivo 7 tesla magnetic resonance imaging.
Versteeg, Guus (2025) Revealing the margins : Towards ex vivo image-guided surgery in oral cancer resection : Facilitating intraoperative assessment of resection margins during oral squamous cell carcinoma surgery using automatic segmentation and 3D visualisation on ex vivo 7 tesla magnetic resonance imaging.
|
PDF
3MB |
| Abstract: | Rationale The 7Tex study at the UMC Utrecht aims to enable intraoperative assessment of resection margins (IOARM) during oral squamous cell carcinoma (OSCC) surgery within 45 minutes, using 7 tesla magnetic resonance (MR) imaging to reduce the incidence of inadequate margins. To facilitate this, an automatic pipeline for tumour segmentation and intuitive resection margin visualisation is required. Methods 55 transverse and 47 sagittal T2-weighted (T2w) MR scans of resection specimens were used to train deep-learning networks for automatic segmentation of tumour tissue. The generation of an intuitive, three-dimensional (3D) resection margin visualisation model was achieved using Python scripting in dedicated imaging software. This 3D model was evaluated by all surgeons performing OSCC resections at the UMC Utrecht using the System Usability Scale (SUS). Feasibility of super-resolution algorithms was explored by reconstructing coronal T2w scans using only the transverse and sagittal T2w scans. Results The networks achieved Dice similarity coefficient scores of 0.695 ± 0.231 (transverse) and 0.698 ± 0.164 (sagittal). The 3D model could be generated within five minutes of the specimen scan and tumour segmentation being provided. Usability of the 3D model was positively evaluated by all surgeons, with an average SUS score of 74.4. Super-resolution reconstruction returned an average structural similarity index measure of 0.766 ± 0.0842 and a root mean squared error of 4.16% ± 1.54 relative to total signal intensity. Conclusion This study presents a clinical pipeline which facilitates IOARM in OSCC surgery, combining automatic tumour segmentation and intuitive 3D visualisation of ex vivo 7T MR scans. While some logistical challenges remain to be addressed, the technical foundation for clinical implementation of the pipeline is now established. |
| Item Type: | Essay (Master) |
| Faculty: | TNW: Science and Technology |
| Subject: | 44 medicine, 50 technical science in general |
| Programme: | Technical Medicine MSc (60033) |
| Link to this item: | https://purl.utwente.nl/essays/107843 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page