Mise en place d'une méthode de radiothérapie adaptative à l'aide de l'imageur portal
|Advisor:||Archambault, Louis; Varfalvy, Nicolas|
|Abstract:||This thesis project aims to improve cancer treatments by external radiotherapy using images from the electronic portal imaging device (EPID). The portal imager is a detector, attached to the gantry on the opposite side of the radiation source and behind the treatment table when a patient is treated. This enables the collection of the signal from the accelerator, with or without the patient on the treatment table. Currently, the total dose of a curative treatment by external radiotherapy is delivered over several weeks. During these weeks, the patients may have anatomical changes due to multiple factors (such as weight loss or changes in tumor volume), which may compromise the accuracy of the delivered dose as compared to the planned treatment. The aim of this research project is to develop a new adaptive radiotherapy methodology allowing to determine at each stage of the treatment whether the patient’s anatomy is still consistent with the initial planning and to estimate the dosimetric impact when changes were observed. EPID dose images were acquired at each treatment fraction for the three different anatomical site under study (head and neck, prostate and lung), together with cone beam volumetric imaging acquisition taken at least twice during treatment, at the second and last fractions. An image extraction and analysis software was first developed to facilitate monitoring and eventually to include the process in an automated clinical routine. A gamma analysis was thus performed for each image, with the one from the first fraction set as reference and several parameters were extracted from these analyzes: average γ-value, standard deviation, Top 1% from maximum γ, the percentage of points with a γ value larger than 1 and the largest connected pixel area with a gamma value > 1. Analyses were made with 3%/3mm as passing criteria and with a threshold of 10% for the low value cut-off. Since each targeted anatomical site has its own characteristics, they were all assessed independently. Taking into account the constraints of each anatomic site, the specificity and the sensitivity values that we have determined, confirmed the performance of the method to detect anatomical changes. Moreover, the γ-analysis threshold correlated well with morphological changes having a relevant dosimetric impact. In conclusion, such analyses of daily EPID images allow to identify patients at risk of deviation from the planned treatment and can support an early replanning decision. Our method is very quick to implement and very easy to use, and should be a powerful tool to improve the treatment of cancers, particularly those of the three anatomical sites studied|
|Document Type:||Thèse de doctorat|
|Open Access Date:||17 April 2019|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.