University of Twente Student Theses

Login

MR image biomarkers of the parotid gland stem cell rich region to predict xerostomia after radiotherapy

Vette, S.P.M. de (2021) MR image biomarkers of the parotid gland stem cell rich region to predict xerostomia after radiotherapy.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Embargo date:22 December 2022
Abstract:Purpose - To evaluate the addition of Magnetic Resonance Imaging (MRI) image biomarkers (IBMs) of the parotid gland stem cell rich (SCR) region to a clinical reference model to predict daytime xerostomia 12 months after radiotherapy for head and neck cancer (HNC). Methods and Materials - A retrospective analysis was performed on 104 HNC patients who were treated using curative radiotherapy between 2018 and 2020. T1 turbo spin echo MRI scans, planning Computed Tomography (CT) scans, dose distributions and GRIX daytime xerostomia grade (baseline and 12 months follow-up) were prospectively collected. SCR regions were delineated on CT. Parotid gland and SCR region structures were transferred to MRI scans, MRI scans were normalized (standardization by intensities), and IBMs were extracted from the ipsilateral and contralateral SCR region. Dose parameters were determined based on the dose distribution. Pre-selection of IBMs was executed using the Bayesian Information Criterion and Spearman correlation with the endpoint (<0.8). Logistic regression sub-models were created based on predictor groups with intercorrelation of <0.8 and sub-models were combined in a composite model. Internal validation was executed by bootstrapping 100 times. Selected IBMs were added to a clinical reference model and using a likelihood-ratio test, the addition of IBMs to the model was tested. Results - Predictive IBMs were the long run low grey level emphasis ipsilateral and the short run high grey level emphasis contralateral. The area under the curve (AUC) for this model was 0.68 (0.43-0.86). After internal validation, the AUC decreased to 0.57 (0.45-0.68). The clinical reference model had an AUC of 0.55 on our study population after closed-testing procedure and the likelihood-ratio test revealed that adding the IBMs did not improve this model. Discussion - With another MRI sequence specifically created for sialography and subsequent a segmentation of the SCR region based on the main ducts of the parotid gland, the SCR region could be defined more accurately, most likely resulting in more predictive IBMs. Furthermore, due to differences in the study population, the clinical reference model did not have a good fit on our study population. These differences were probably caused by changes in treatment planning and technological progress over the years. In conclusion, it was found that MRI IBMs of the SCR region can be predictive of daytime xerostomia 12 months after radiotherapy, but more research is needed.
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/89184
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page