University of Twente Student Theses


Optimization of 3D MRI topography for stereotactic navigation in pelvic surgery

Schram, Renske (2023) Optimization of 3D MRI topography for stereotactic navigation in pelvic surgery.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:Using stereotactic navigation in surgery for locally advanced and recurrent rectal cancer has proven advantageous. However, the segmentation of the tumor and vital structures is a time-consuming process. Therefore, the objective of this thesis was to enhance the preoperative workflow and present several implementations to optimize the extracting of patient-specific 3D topographies from standard pelvic and abdominal MRI and CT scans, emphasizing three primary elements: optimization of the use of standard protocol imaging, automated segmentation, and multi-model registration for pelvic 3D-topography mapping. The dataset optimization involves isotropic volumetric reconstruction for standard protocol anisotropic MRI scans from various scanning directions. This proposed method significantly enhances the identification and continuous recognition of extrasacral parts of the sacral nerves and internal iliac arteries without compromising the quality of other characteristics. These improvements indicate the potential for reconstructed MRI volumes to enhance the validation and manual adjustment processes and the feasibility of creating a single isotropic high-resolution volume for 3D-topography segmentation and mapping. To complement existing tools and research for automated segmentation of pelvic structures and reduce the time required for segmenting under-explored structures for pelvic 3D-topography, the sacral nerves and piriformis muscle, two deep learning methods were explored: the use of nnU-net and semi-automated pseudo RGB-patched transfer learning. Quantitative evaluation of the segmentation performance indicates substantial time-saving benefits, with the use of 3D nnU-net suggested as the most promising model. For the deformation of the MRI-based delineations to the CT imaging domain for navigational implementation, ANTs SyN registration is proposed to improve the multi-modal registration in the pre-operative workflow for 3D topography mapping. Visual evaluation of the resulting 3D-topographies and delineation deformation accuracy indicates a tendency toward clinical benefits when using the reconstructed MRI scans as the foundational imaging for extracting 3D-topography. While each element can be optimized individually, it is essential for an efficient workflow that these elements are effectively connected and integrated correctly. This thesis's main contribution lies in optimizing these elements and evaluating the clinical impact individually and collectively, specifically tailored pelvic 3D topography mapping. Consequently, this thesis presents the initial steps towards an optimized pre-operative workflow to improve the clinical feasibility of stereotactic navigation for rectal surgery.
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:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page