Ces dernières années, les techniques de traitement des cancers par radiothérapie externe se sont complexifiéesafin de cibler la tumeur tout en protégeant les organes à risque. Les systèmes de planification de traitement (TPS)réalisent un calcul prévisionnel de la distribution de la dose absorbée dans le patient (via des images CT).Afin d’obtenir un résultat de dose précis dans un temps raisonnable, le calcul est effectué par des algorithmessimplifiés. En présence de dispositifs médicaux métalliques de masses volumiques élevées (prothèses de hancheou prothèses dentaires), les algorithmes atteignent leurs limites. De plus, ces dispositifs perturbent lareconstruction tomodensitométrique en créant des artéfacts sur les images rendant difficile la délinéation desorganes. L’objectif de ce travail a été d’évaluer les algorithmes implémentés dans les TPS en présenced’hétérogénéités de forte densité avec des mesures expérimentales et le code de calcul Monte-Carlo BEAMnrcdans un fantôme anthropomorphique tout d’abord avec des échantillons naturels, puis avec des échantillonscalibrés. Ensuite, une évaluation rétrospective des algorithmes cliniques par rapport à Monte-Carlo a été réaliséeavec des patients traités en RC3D et en RCMI.Les mesures ont mis en évidence une atténuation pouvant aller jusqu’à 17 % pour l’amalgame dentaire parrapport à l’algorithme clinique sur les images CT avec artéfacts, créant une zone de sous-dosage dans le volumecible. L’ensemble des résultats a donné lieu à des recommandations pour la clinique (corriger les images CT si levolume cible est à moins de 3 cm d’une prothèse, privilégier l’algorithme AAA plutôt que Pencil Beam…). / The last few years, cancer treatment techniques in radiation therapy have become more complex to better targetthe tumor while protecting the organs at risk. The treatment planning systems (TPS) achieve a predictivecalculation of the distribution of the dose absorbed by the patient (via CT images).In order to obtain an accurate dose result within a reasonable time, the calculation is performed with simplifiedalgorithms. In the presence of medical devices made of high density metal (hip prosthesis or dental prosthesis),the algorithms reach their limits. Moreover, these devices disrupt computed tomography reconstruction, creatingartifacts on the images and thus making difficult the delineation of organs. The aim of this work is to evaluatethe algorithms of the TPS in the presence of high density heterogeneity using experimental measurements andthe Monte Carlo BEAMnrc code in an anthropomorphic phantom: on one hand with natural samples, and on theother hand, with calibrated samples. Then, a retrospective evaluation of clinical algorithms compared to MonteCarlo is achieved using treated patients in Conformal Radiotherapy and in Intensity Modulated RadiationTherapy (IMRT). The measurements show an attenuation of up to 17 % for dental amalgam compared with theclinical algorithm on CT images with artifacts, creating an under-dosage area in the target volume. All theseresults lead to recommendations for the clinical treatments (corrected CT images if the target volume is closerthan 3 cm to prosthesis, favor the AAA algorithm rather than Pencil Beam …).

<p>Cone beam computed tomography (CBCT) is a 3D x-ray imaging technique in which the x-ray beam is transmitted to an object with wide beam geometry producing a 2D image per projection. Due to its faster image acquisition time, wide coverage length per scan, and fewer motion artifacts, the CBCT system is rapidly replacing the conventional CT system and becoming popular in diagnostic and therapeutic radiology. However, there are few studies performed in CBCT dosimetry because of the absence of a standard dosimetric protocol for CBCT. Computed tomography dose index (CTDI), a standardized metric in conventional CT dosimetry, or direct organ dose measurements have been limitedly used in the CBCT dosimetry.</p> <p>This dissertation investigated the CBCT dosimetry from the CTDI method to the organ, effective dose, risk estimations with physical measurements and Monte Carlo (MC) simulations.</p> <p>An On-Board Imager (OBI, Varian Medical Systems, Palo Alto, CA) was used to perform old and new CBCT scan protocols. The new CBCT protocols introduced both partial and full angle scan modes while the old CBCT protocols only used the full angle mode. A metal-oxide-semiconductor-field-effect transistor (MOSFET) and an ion chamber were employed to measure the cone beam CTDI (CTDI_CB) in CT phantoms and organ dose in a 5-year-old pediatric anthropomorphic phantom. Radiochromic film was also employed to measure the axial dose profiles. A point dose method was used in the CTDI estimation.</p> <p>The BEAMnrc/EGSnrc MC system was used to simulate the CBCT scans; the MC model of the OBI x-ray tube was built into the system and validated by measurements characterizing the cone beam quality in the aspects of the x-ray spectrum, half value layer (HVL) and dose profiles for both full-fan and half-fan modes. Using the validated MC model, CTDI_CB, dose profile integral (DPI), cone beam dose length product (DLP_CB), and organ doses were calculated with voxelized MC CT phantoms or anthropomorphic phantoms. Effective dose and radiation risks were estimated from the organ dose results.</p> <p>The CTDI_CB of the old protocols were found to be 84 and 45 mGy for standard dose, head and body protocols. The CTDI_CB of the new protocols were found to be 6.0, 3.2, 29.0, 25.4, 23.8, and 7.7 mGy for the standard dose head, low dose head, high quality head, pelvis, pelvis spotlight, and low dose thorax protocols respectively. The new scan protocols were found to be advantageous in reducing the patient dose while offering acceptable image quality.</p> <p>The mean effective dose (ED) was found to be 37.8 ±0.7 mSv for the standard head and 8.1±0.2 mSv for the low dose head protocols (old) in the 5-year-old phantom. The lifetime attributable risk (LAR) of cancer incidence ranged from 23 to 144 cases per 100,000 exposed persons for the standard-dose mode and from five to 31 cases per 100,000 exposed persons for the low-dose mode. The relative risk (RR) of cancer incidence ranged from 1.003 to 1.054 for the standard-dose mode and from 1.001 to 1.012 for the low-dose mode.</p> <p>The MC method successfully estimated the CTDI_CB, organ and effective dose despite the heavy calculation time. The point dose method was found to be capable of estimating the CBCT dose with reasonable accuracy in the clinical environment.</p> / Dissertation

The ability to accurately predict dose from electron backscatter created by internal lead shielding utilized during various superficial electron beam treatments (EBT), such as lip carcinoma, is required to avoid the possibility of an overdose. Methods for predicting this dose include the use of empirical equations or physically measuring the electron backscatter factor (EBF) and upstream electron backscatter intensity (EBI). The EBF and upstream EBI are defined as the ratio of dose at, or upstream, from the shielding interface with and without the shielding present respectively. The accuracy of these equations for the local treatment machines was recognised as an area that required verification; in addition the ability of XiO's electron Monte Carlo (eMC) treatment planning algorithm to handle lead interfaces was examined. A Monte Carlo simulation using the EGSnrc package of a Siemens Artiste Linac was developed for 6, 9, 12, and 15 MeV electron energies and was verified against physical measurements to within an accuracy of 2 % and 2 mm. Electron backscatter dose distributions were predicated using the MC model, Gafchromic film, and XiO eMC, which when compared showed that XiO's eMC could not accurately calculate dose at the lead interface. Several MC simulations of lead interfaces at different depths, corresponding to energies of 0.2-14 MeV at the interfaces, were used to validate the accuracy of the equations, with the results concluding that the equation could not accurately predict EBF and EBI values, especially at low energies. From this data, an equation was derived to allow estimation of the EBF and upstream EBI, which agreed to within 1.3 % for the EBF values and can predict the upstream EBI to a clinically acceptable level for all energies.

<p>The main focus of this work is to improve the existing clinical machine model within the Pinnacle software planning system (at Juravinski Cancer Center, Hamilton, CA). The incident energy fluence spectrum exiting from the accelerator head is considered an important element of the machine model. Relying on the Pinnacle auto modeling function to determine the relative photon fluence spectrum based on percent depth dose curves fitting for various filed sizes, led to different solutions when the process cycle were repeated. This work presents a new method for determining the Pinnacle photon energy fluence spectrum based on 6 MV Varian 21EX machine. A Monte Carlo simulation spectrum based on BEAMnrc code was attenuated to various depths of water. We determine that, the BEAMnrc spectrum attenuated by 15 cm of water gives the closest agreement between the computed and measured depth dose, similar to the clinical machine spectrum.</p> <p>Implementing the novel spectrum into a machine that retained the same modeling parameters as the clinical machine (21ex-JCC) shows a slight better calculation of the output factor. The MLC model parameters were also investigated, however, adjusting the MLC offset table was found to give significant improvements, especially for the small field geometries.</p> <p>The full impact of adjusting the photon energy spectrum, Off-Axis Softening Factor, MLC rounded leaf tip radius and MLC calibration offsets were investigated individually, resulting in a good model parameter fit. Several proposed supplementary setups were created to further assess our model. This include a geometry sensitive to MLC abutment leakage, the calculation of output factors for long and narrow MLC defined fields, and small square MLC and jaws defined fields. A Sun-Point diode detector was used in the measurement of the output factors for its accurate precision at small geometries. In addition, a GAFCHROMIC EBT2 film dosimetry was used in the measurement of the MLC abutment leakage.</p> <p>Our new model shows superior results in comparison to the clinical 21ex-JCC machine model, especially with MLC small field calculations. We conclude that relying on PDD curves and dose profiles validation method in assessing the model might not necessarily lead to the best machine parameters, since these are not sensitive to subtle changes in parameters that have important dosimetric consequences.</p> / Master of Science (MSc)