University of Twente Student Theses


Perfusion of the abdominal aortic aneurysm wall with dynamic contrast enhanced MRI : a potential biomarker for aneurysm progression and rupture?

Rijken, L. (2022) Perfusion of the abdominal aortic aneurysm wall with dynamic contrast enhanced MRI : a potential biomarker for aneurysm progression and rupture?

[img] PDF
Full Text Status:Access to this publication is restricted
Embargo date:30 May 2023
Abstract:BACKGROUND – Nowadays, the maximal diameter is used as a measure for the treatment of abdominal aortic aneurysms (AAAs). However, the predictive value of the maximal diameter for aneurysm rupture is not always sufficient. Hence, there is a need for a better patient-individual prediction of AAA progression and risk of rupture. The perfusion of the AAA wall may be a potential biomarker. AAA wall perfusion can be measured using Dynamic Contrast Enhanced MRI (DCE-MRI) by the quantification of the volume transfer constant Ktrans. METHODS – 3D DCE-MRI data of twenty AAA patients were analyzed. Seventeen patients received two consecutive DCE-MRI scans at a one-week interval. Two different methods were assessed to perform pharmacokinetic modeling for generating Ktrans maps. The conventional least squared (LSQ) fitting method was compared to a physics-informed deep learning network (DCE-net). Each method was evaluated for reproducibility of scan and rescan. Subsequently, clinical analysis of the generated 3D Ktrans maps of the aneurysm wall was performed. AAA diameter and growth per year were correlated to median and maximum Ktrans values. RESULTS – In total, seventeen patients underwent consecutive DCE-MRI scans and were included for further Ktrans analysis. The Ktrans maps were moderately reproducible in both the LSQ fitting and DCE-net methods. The coefficients of variation (CoV) of the LSQ fitting and the DCE-net for the small segments were 45.4% and 48.5%, respectively, and for large segments 24.5% and 27.3%, respectively. While comparing the small and large regions of the 3D Ktrans maps, there was a fixed bias between scan and rescan (-0.05 min-1 for the LSQ fitting and -0.04 min-1 for the DCE-net). There was no fixed bias of median Ktrans values of the whole AAA wall between the two scans. Clinical analysis was performed for the Ktrans maps generated with the LSQ method. Median Ktrans of the whole annotated AAA of the first DCE-MRI scan and the AAA diameter measured on ultrasound had a significant correlation (R=-0.51, P = 0.043). There was a moderate correlation between the 95th percentile Ktrans value of the whole annotated AAA of the first scan and the mean growth per year (R=- 0.51, P = 0.045). CONCLUSION – In this study, Ktrans could be quantified on 3D DCE-MRI data using the LSQ fitting and DCE-net. The DCE-net did not outperform the conventional LSQ fitting. Still, the derived parameter maps from DCE-MRI data were sensitive to acquisition and post-processing techniques. The generated 3D Ktrans maps were more robust for larger regions compared to smaller regions. Furthermore, the clinical use of Ktrans as a biomarker for progression and rupture could not yet be determined based on the results of this thesis.
Item Type:Essay (Master)
Amsterdam UMC, Amsterdam, Netherlands
Faculty:TNW: Science and Technology
Subject:44 medicine
Programme:Technical Medicine MSc (60033)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page