Javascript must be enabled for the correct page display

Dosimetric validation of sCT images created from CBCT and MR images by a DCNN for the use of adaptive proton therapy in head and neck cancer patients.

de Jong, Bas (2019) Dosimetric validation of sCT images created from CBCT and MR images by a DCNN for the use of adaptive proton therapy in head and neck cancer patients. Master's Thesis / Essay, Biomedical Engineering.

[img]
Preview
Text
mBME_2019_B.A.deJong.pdf

Download (2MB) | Preview
[img] Text
toestemming.pdf
Restricted to Registered users only

Download (212kB)

Abstract

Abstract Adaptive proton therapy refers to the possibility to adjust treatment plans during the course of fractionated treatment to compensate for changes in patient anatomy. This insures sufficient treatment of the tumor while minimizing damage to surrounding organs at risk (OARs). In current clinical practice, adaptive proton therapy relies on computed tomography (CT) images. CT images are linked to imaging dose and thus the imaging frequency has to be well balanced. Cone beam computed tomography (CBCT) and magnetic resonance (MR) imaging are an al- ternative to CT imaging and can give a daily representation of the patient anatomy. However these images are not directly suited for proton dose calculations. In this work, a Deep Convolutional Neural Network was trained and used to create synthetic CT (sCT) images based on CBCT and MR images. The aim of this research is to evaluate and compare the accuracy of proton dose distributions calculated on sCT images based on CBCT and MR images by a Deep Convolutional Neural Network (DCNN), for the use of adaptive proton therapy in head and neck cancer patients. The quality of the sCT images was assessed by calculating mean absolute error (MAE) values for the sCT images, with respect to corresponding CT images. Geo- metric accuracy of the reconstruction of bony structures is assessed using the Dice similarity coefficient (DSC). Dose distributions were calculated on sCT images and corresponding CT images using clinical treatment plans for 9 head and neck cancer patients. The accuracy of the proton dose calculations on sCT and corresponding CT images is compared and evaluated using gamma analysis and evaluation of the dose in clinically delineated organs at risk (OARs). Average MAE values were found of 37 ± 4 HU and 58 ± 4 HU respectively for sCT images based on CBCT and MR images respectively. The Gamma analysis resulted in average pass rates of 98.6 ± 1% and 97.5 ± 1% for dose distributions calculated on sCT images created from CBCT and MR images, respectively. Fur- thermore, evaluation of dose volume histograms for the planning treatment volume (PTV) and OARs, showed that sCT images based on CBCT as well as on MR im- ages are suitable for proton dose calculations, with avarage dose differences of less than 1%. From the results of this research we conclude that both sCT images based on CBCT and MR images could be suitable for the use in adaptive proton therapy for head and neck patients. The use of sCT images for adaptive proton therapy enables daily evaluation of the impact of anatomical changes on the treatment. Furthermore, it simplifies workflows making the acquisition of repeated CTs (rCTs) along the treatment re-dundant.

Item Type: Thesis (Master's Thesis / Essay)
Supervisor:
Supervisor nameSupervisor E mail
Knopf, A.C.a.c.knopf@umcg.nl
Degree programme: Biomedical Engineering
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 13 Aug 2019
Last Modified: 13 Aug 2019 10:03
URI: http://fse.studenttheses.ub.rug.nl/id/eprint/20650

Actions (login required)

View Item View Item