The objective of this work is to use MOSFET dosimeters to accurately measure surface dose delivered during CT examinations in various scanning conditions. To achieve this, the behaviour of MOSFETs under kilovoltage x-ray irradiation first needed to be investigated. A dose-to-dose reproducibility of 4.5%, and a mean change in sensitivity response of 10.4% with accumulated dose were measured. A Monte Carlo model of the x-ray source of a PQ5000 CT simulator was built and validated in order to investigate the MOSFET response characteristics and perform dose calculations. An over-response of 10% was observed when the beam energy was decreased from 140 to 80 kVp, and a slight anisotropy of 8.5% from the mean value over 360º was observed. The dosimeters were calibrated on a solid water phantom using a method involving MC surface dose calculations. Good agreement was found between measurements and simulations of surface dose on a cylindrical PMMA phantom for a stationary tube technique, single axial scan and multiple contiguous axial scans, with generally less than 7.5% discrepancies. Film and MOSFET measurements were then performed for helical adult brain scan parameters using different pitch and collimator settings. The use of five MOSFETs combined in a linear array was found to be suitable to accurately measure surface dose in helical scans for almost all pitch and collimation combinations.

Since portal images are created by megavoltage, forward-directed bremsstrahlung beams, their image quality is inferior to that of images produced by kilovoltage beams. In this study, characteristics of orthogonal bremsstrahlung photons produced by megavoltage electron beams were studied and their suitability for radiotherapy imaging was evaluated. Orthogonal bremsstrahlung beams with kilovoltage effective energies can be obtained from megavoltage electrons striking low atomic number targets. A 10 MeV electron beam emerging out of the research port of a Varian Clinac-18 linac was made to strike carbon, aluminum and copper targets. Percentage depth dose and attenuation measurements of forward and orthogonal beams were performed, and experimental results were compared with Monte Carlo-calculated findings. Images of simple contrast objects taken using the orthogonal bremsstrahlung beams showed superior contrast levels in comparison to those produced by the forward beams.

Dynamic beam delivery (DBD), either in the step-and-shoot or dynamic mode, promises to be useful for developing conformal radiotherapy techniques. The introduction of DBD in the clinic requires stringent evaluation of the DBD system before any dynamic technique can be performed clinically. In this work, a series of quality assurance tests were performed to evaluate a DBD system (DBD Toolbox) on a linear accelerator. The results of the tests have shown that the DBD Toolbox is capable of precisely controlling jaw motion, collimator rotation, and gantry rotation for dynamic beam delivery to within a 1 mm tolerance for jaw motion and 0.5° tolerance for collimator or gantry rotation. In addition, a simple model for dynamic jaw dosimetry was evaluated. The quality assurance tests and model evaluation are presented in this thesis.

This thesis addresses several brain MRI segmentation methods including three methods of using normal distributions, similar methods using t-distributions and a new method using MCLUST software packages. The main purpose of this thesis is to try to improve current brain image segmentation methods and reduce the long computing time with satisfactory results. The methodology consists in applying histogram analysis for the initialization of parameters, using Simpson's rule for the approximation of numerical integration and describing methods of skull stripping and neighborhood information encoding. The two most significant contributions are suggestions for speeding up current methods and a newly proposed method that ignores spatial information for parameter estimation. Potential future work is proposed at the end.

Accurate dose calculations and analysis tools are essential to radiotherapy treatment planning. Radiotherapy deliveries utilize the information provided by the treatment planning system and it is generally accepted that clinical outcome can be improved if accuracy in the dose delivery is further improved. Proven Monte Carlo calculations increase the planning accuracy however, most radiotherapy departments do not use Monte Carlo. The McGill Monte Carlo treatment planning system, MMCTP, provides a flexible software environment to integrate Monte Carlo planning with current and new treatment modalities and deliveries. The MMCTP design consists of a graphical user interface, which runs on a simple workstation connected through standard secure-shell protocol to a cluster for lengthy Monte Carlo calculations. The impact of this tool lies in the fact that it allows for systematic, platform independent, large-scale Monte Carlo planning calculations for different treatment sites. Various measurements and patient recalculations were preformed to validate the software and ensure proper functionality.

The rotational total skin electron irradiation technique at the McGill University Health Centre has undergone several developments over the past few years. Replacement of the formerly used linear accelerator has prompted many modifications to the previously reported technique. With the current technique, the patient is treated while standing on a rotating platform, with a single large field at a source to surface distance of 378 cm. The electron field is produced by a Varian 21EX linear accelerator using the commercially developed 6 MeV high dose rate total skin electron mode, along with a custom-built flattening filter. Ionization chambers, radiochromic film, and MOSFET detectors have all been used to confirm the dosimetric properties of this technique. Measurements investigating the stationary beam properties, the effects of full rotation, and the dose distributions to a humanoid phantom are reported. In addition, comparisons with commonly-used stationary beam treatment alternatives are presented.

The thesis work contains two related projects. The first project examines the characteristics of the Farmer-type ionization chamber (Farmer-type IC 2571 A) used in our experiments. An investigation was carried out to determine if the simplified "two-voltage technique" used in clinics is sufficiently accurate for saturation current estimation. Three models were used to fit the measured currents and to estimate the saturation current. The estimation was carried out by measuring the currents set-up in the Farmer-type IC while varying the applied voltages, for both polarities, for fixed and variable dose rates. The second project investigated peripheral dose in 6 MV beam using the Farmer-type IC. The signal produced in the ionization chamber when measuring the peripheral dose is 2 to 3 orders of magnitude lower than the signal produced with the chamber in the primary beam, thus, the leakage and saturation characteristics of the ionization chamber had to be investigated to ensure that they do not have adverse effects on the chamber reading. Measurements were acquired in open field (10x10 cm2) and in a dynamic MLC field (10x1 cm2), with the IC at the machine's isocenter. / Measurements allowed us to determine peripheral dose due to scatter and leakage radiation. Measurements were repeated for a blocked beam, which allow us to determine only the leakage radiation component of the peripheral dose. The two components of the peripheral dose depend strongly on the linac head configuration and shielding. Leakage radiation per MU for a dMLC field is similar or higher than the leakage radiation for a static field, which implies an increase in peripheral dose for IMRT-type treatments. Knowledge of peripheral dose can be useful in estimating dose and risk to sensitive structures outside of the primary treatment field.

Electron arc therapy is a special radiotherapeutic technique using a rotational electron beam in the treatment of large superficial tumours following curved surfaces. In those cases, arc therapy offers the best way to optimize dose uniformity while sparing healthy tissues and critical organs. The use of this technique overcomes under or over dosage problems caused by field junctions. However, electron arc therapy presents important challenges in terms of dosimetry and treatment planning. / Clinical implementation of electron arc therapy requires the study of many parameters of influence such as the radius of curvature of the treated surface, the width of the treatment field, the total angle of irradiation and the beam energy. Monitor unit calculation to deliver prescribed dose is a very critical topic and, in general, requires acquisition of a large amount of measured dosimetric data. / This project concerns the clinical implementation of electron arc therapy using an Elekta SL-25 linear accelerator in the radiation oncology department of the Maisonneuve-Rosemont Hospital (Montreal, Canada). Firstly, the objective of the study is to observe the influence of the radius of curvature, the total arc angle and the field width on the following dosimetric parameters: depth of maximum dose, isodose distributions and electron arc beam output at the depth of maximum dose. Secondly, for our particular setup, the goal is to develop a simple monitor unit calculation method, based on an analytical model fitted through measured dosimetric data covering a large range of possible clinical situations. / In order to achieve these goals, electron arc irradiations were performed on cylindrical acrylic phantoms of different radii, successively varying the total arc angle and the field width at isocentre. Results obtained with thermoluminescent dosimeters show a minor impact of the radius of curvature variation on the percent depth dose curves. However, they show a significant impact on the beam output. It was also observed that the total arc angle influences the dose at the depth of maximum dose only up to a limit angle value, different for each radius of curvature. Finally, the field width at isocentre has an impact on the beam output as well as on the bremsstrahlung contribution at the isocentre. / Concerning the monitor unit calculation, a seven parameter analytical model fitted through measured data was obtained using Origin 7 software. A relationship giving the beam output as a function of the radius of curvature and the total arc angle was found. The field width was not included in the model, but will be part of further investigation before clinical implementation. As future work, dosimetric measurements with other energies should be carried on, mainly to be able to cover a wider range of clinical cases.

Measurements of absorbed dose to water in megavoltage electron beams are carried out with cylindrical or plane-parallel ionization chambers. The perturbation factor associated with plane-parallel ionization chambers that have sufficiently large guard rings is assumed to be unity. The NACP-02 chamber is one such chamber that we have investigated in order to determine if this assumption is valid. By performing Monte Carlo simulations with the EGSnrc code, water-to-air Spencer-Attix stopping powers and perturbation factors were calculated in water with a detailed model of the NACP chamber that was validated through results obtained from measurements and simulations. These were determined for electron beam energies between 4 MeV and 19 MeV for a clinical and PSDL linear accelerator. The overall perturbation factor was found to decrease with energy at dref from 1.015 at 4 MeV to 1.004 at 19 MeV and increase with depth for a fixed beam quality.

A three dimensional, automatic, anatomy-based system for portal verification has been developed based on an FFT implementation of Pearson's correlation coefficient (PCC). The PCC requires no anatomy or point-pair identification, is robust when encountering changes in scaling and shifts in image amplitudes and requires no priori knowledge of the anatomy, which makes it an ideal candidate for portal-to-DRR image registration. Features for matching are selected from orthogonal portal images and compared to the corresponding megavoltage DRR. The position of the highest correlation value is then converted into beam-to-patient geometry and compared with the actual patient setup. By continuously generating DRRs, the system is capable of verifying translation errors, in-plane rotation and out-of-plane rotation errors. The mean accuracy of translation and rotation registrations tests were 0.58 mm and 0.79° respectively for DRR-to-DRR matching, and 1.22 mm and 1.31° respectively for portal-to-DRR matching.