Introduction Cerebral glioma is the most prevalent primary brain tumour, of which the majority are high grade gliomas. High grade gliomas possess a poor prognosis, and glioblastoma patients survive less than one year after diagnosis. To date, histological grading is used as the standard technique for diagnosis and survival prediction. Previous studies using advanced techniques such as MR Perfusion have achieved a high sensitivity but a low specificity in identifying high grade gliomas. Moreover, they have failed to distinguish glioblastoma from anaplastic glioma. The purpose of the study presented here is to assess the diagnostic and prognostic value for cerebral glioma of cerebral blood volume maps derived from MR perfusion. Methods This retrospective study was approved by the local research ethics committee and clinical audit office. This study included 123 patients with newly diagnosed cerebral glioma, of all grades. Histological diagnosis was used as the standard reference for all potential patients. The relative tumour blood volume (rTBVmax) derived from MR perfusion was used for radiological grading of cerebral glioma. Receiver operating characteristics (ROC) were used to define the best threshold value in distinguishing the glioma grades and in determining the accuracy values (sensitivity, specificity, and positive and negative predictive values). For survival analysis, Kaplan-Meier was used to illustrate and compare the discriminatory value of the histological and radiological classifications. A multiple Cox regression model was used to assess the prognostic value of both classifications in addition to other tested demographic and clinical variables. Finally, the influence of potential moderators was assessed using ANOVA, to assess whether the variation in rTBVmax was only due to the difference in tumour grades. Results A model data set (n = 50) produced a 7-fold increase of TBVmax in tumour versus white matter and provided sensitivity and specificity of 97% and 94%, respectively, in distinguishing high versus low grade glioma. Moreover, a threshold value of 9.6 provided sensitivity and specificity of 100% and 56% in differentiating glioblastoma within the group of high grade gliomas. These threshold values were applied to the second group (n = 73) and provided sensitivity and specificity of 96% and 95% in distinguishing high versus low grade glioma, and 97% and 73% in differentiating, within the high grade gliomas, glioblastoma from anaplastic glioma. Using these two thresholds for a three-tier radiological classification, both the Kaplan-Meier plots and the multiple Cox regression showed that radiological classification was the most independent predictor of survival and tumour progression. The proposed radiological classification system was better than histological classification in predicting glioma patients survival especially noted in a group of moderately hyperaemic rTBVmax. Conclusion MR perfusion is a non-invasive and robust technique in glioma grading and survival prediction. The diagnostic value of rTBVmax derived from MR perfusion in differentiating high versus low grade glioma is promising. It may have a role in the future in defining the appropriate treatment. However, the proposed radiological classification was inferior in differentiating anaplastic glioma from glioblastoma multiforme. In the future, a more advanced multimodal MR, such as MR spectroscopy and MR diffusion, may be studied, besides MR perfusion, in order to improve this diagnostic accuracy.

A study is performed to assess the utility of a computer numerically controlled (CNC) mill to produce missing tissue compensating filters and for the delivery of intensity-modulated beams for inverse treatment planning. A computer aided machining (CAM) software is used to assist in the design and construction of such filters. Geometric measurements of stepped and wedged surfaces are made to examine the accuracy of surface milling. Results show that the deviation of the filter surfaces from design does not exceed 1.5%. Effective attenuation coefficients are measured for CadFree and Cerrobend in a 6 MV photon beam. The ability of the CNC mill to accurately produce surfaces is further verified with dose profile measurements in a 6 MV photon beam. Dose profiles, measured beneath the test phantoms and beneath a flat phantom are compared to those produced by a commercial treatment planning system. Agreement between measured and predicted profiles is within 2%, indicating the viability of the system for filter production.

The dosimetric characteristics of a standard Varian 52-leaf multileaf collimator (MLC) and BrainLAB m3 micro-multileaf collimator (micro-MLC) have been investigated for square, rectangular, and irregular fields for 6 MV and 18 MV photon beams provided by a Varian Clinac 2300 C/D linear accelerator (linac). The percentage depth dose data and the conventional collimator factor are unaffected by the addition of MLC or micro-MLC shaped field unless, in the latter case, the tertiary field is much less than the jaw setting. However, relative dose factors for a given MLC or micro-MLC field size depend on the jaw setting. The penumbra is generally sharpest for fields defined by the micro-MLC and the least sharp for fields defined by the MLC. Average transmission values were found to be between 1.5% and 2.5%. Comparison and evaluation of two treatments, one delivered using the MLC and the other using the micro-MLC, for the same radiosurgical target volume are described.

This dissertation describes both methodological developments in quantitative functional magnetic resonance imaging (fMRI) of cerebral oxygen consumption, and the results of experiments using these techniques to elucidate the mechanisms linking focal changes in blood flow and oxygen metabolism. Technical contributions presented include a novel MRI pulse sequence for simultaneously monitoring cerebral blood flow and tissue oxygenation with high signal-to-noise ratio, as well as an experiment automation system permitting complex multiparametric studies to be carried out efficiently in large numbers of subjects. These tools enabled us to make a number of significant neurophysiological discoveries with important implications for the design and interpretation of fMRI experiments. In particular, relative changes in cerebral perfusion and oxygen consumption were found to be coupled in a consistent linear ratio of approximately 2:1, respectively, in human visual cortex. A quantitative model predicting that oxygenation-sensitive MRI signals must be extremely sensitive to departures from this coupling ratio was also introduced, revealing that combined perfusion/oxygenation measurement during graded activation is a powerful tool for studying regulatory relationships between these parameters. Predictions based on this model were in excellent agreement with experimental results, supporting model-derived estimates of oxygen consumption and suggesting that the ∼2:1 coupling discovered in visual cortex is likely to apply in most cortical systems. Finally, important non-linear characteristics of fMRI signal dynamics in human visual cortex were revealed, challenging current models of fMRI transient response.

The treatment head of the Clinac-18 medical linear accelerator was modelled using-the EGS4 Monte Carlo simulation package. Photon-energy spectra for fields ranging from 2 x 2 cm$ sp2$ to 20 x 20 cm$ sp2$ in size were generated and the primary and scatter spectra were analyzed separately. The generated x-ray spectra were used in the calculation of the percent depth dose (PDD) distributions for flattened and unflattened 10 MV x-ray beams in a water phantom at a source-surface distance of 100 cm for the various field sizes. The agreement between calculated and measured depth doses is excellent. / Measurements of the dose in the build-up region show that the depth of dose maximum (d$ sb{max}$) increases with increasing field size for fields up to 5 x 5 cm$ sp2$ for both the flattened and unflattened beams. As the field size is increased beyond 5 x 5 cm$ sp2,$ d$ sb{max}$ decreases with increasing field size for the flattened x-ray beam while remaining nearly constant for the unflattened beam. Additionally, the surface dose of the flattened beam is found to approach that of the unflattened beam for large field sizes. Calculations show that the decrease in d$ sb{max}$ as the field size is increased above 5 x 5 cm$ sp2,$ and the rapid increase in the surface dose for the flattened x-ray beam with increasing field size, are due to the degradation of the flattened-beam parameters caused by low-energy photons produced in the flattening filter.

Attenuation and scatter are limiting factors in image quality and quantitation of organ function by single photon emission computed tomography (SPECT). To correct brain images for attenuation an effective water/tissue attenuation coefficient of 0.12 cm$ sp{-1}$ (at 140 keV) or larger has been recommended in order to compensate for the additional bone (skull) attenuation. / It has been determined that the reconstructed images are overcorrected in the centre by 5%, and the optimum correction occurs for a reduced coefficient of 0.09 cm$ sp{-1}$. The overcorrection is due to increased attenuation at the edges of all projections where the path length through the bone is greater, although the bone also increases the scatter at the projection edges. / A correction scheme which uses effective bone and water coefficients was developed to compensate for the attenuation. Alternatively, prior to attenuation correction, a common scatter correction was found to be effective in explicitly removing the bone and water scatter.

In this thesis, the use of the Fricke-gelatin dosimeter in NMR-based radiation dosimetry is investigated. The relationship between the proton spin-lattice relaxation rate and the absorbed dose for the Fricke-gelatin dosimeter is determined using pulsed NMR at 25 MHz. This relationship is used to calculate the NMR dose sensitivity. Practical considerations of Fricke-gelatin dosimetry, such as the spontaneous conversion of ferrous to ferric ions and the effects of oxygen and sodium chloride on the dose response, are determined. Sensitization of the dosimeter by bubbling with a nitrous oxide/oxygen gas mixture is investigated. The system is modelled assuming a multi-site fast exchange between water bound to gelatin, water bound to ions, and water in the bulk. Preliminary aspects of the model are investigated and compared with experimentally determined data. Finally, the utility of the dosimeter is demonstrated by magnetic resonance imaging a Fricke-gel irradiated with an $ sp{192}$Ir radioactive seed.

Three-dimensional (3D) ultrasound is an emerging new technology with numerous clinical applications like measuring the volume of the prostate, monitoring fetal development, or evaluating brain shift during neurosurgery. Ultrasound probe calibration is an obligatory step in order to build 3D volumes from 2D images acquired in a freehand ultrasound system. The role of calibration is to find the transformation that relates the image plane to a sensor attached on the probe. This thesis is a comprehensive review of what has been published in the field of ultrasound probe calibration for 3D ultrasound. The thesis covers the topics of tracking technologies, ultrasound image acquisition, phantom design, speed of sound issues, feature extraction, least-squares minimization, temporal calibration, calibration evaluation techniques and phantom comparisons. The calibration phantoms and methods have also been classified in tables to give a better overview of the existing methods.

Three-dimensional modeling has become an important tool for non-destructive evaluation of tissue for experimental and clinical uses. However, its range of applications is limited by its dependence on expensive imaging modalities. The current research explores ordinary radiographs as an alternative for three-dimensional modeling in several applications relating to vertebral fracture detection and treatment. An algebraic reconstruction algorithm was developed to replace CT in the construction of patient-specific vertebral finite element models for the evaluation of fracture risk in elderly individuals. CT-based models have shown far stronger specificity in detecting at-risk osteoporotic vertebrae than DEXA, but have not been adopted into the clinic because of equipment cost, limited reimbursement codes, radiation exposure, and expertise requirements. Our radiograph-based technique was able to replicate biomechanical predictions made from CT scans to within 7% error, providing an improved alternative to DEXA with less interaction required from the operator than traditional modeling. A similar algebraic technique was employed to investigate the internal changes in a vertebral body during compression to fracture. Because testing apparatus are not compatible with traditional 3-D imaging modalities, we evaluated regional changes in apparent density and deformation of the vertebral shell using periodic sets of 4 radiographs. Damage was observable in an internal area of high strain energy density, then propagated across the inferior endplate causing whole bone failure. Vertebroplasty is the most effective treatment of vertebral compression fractures. It is performed with a minimum of planning and evaluation primarily to avoid cement leakage. However, optimization is required to reduce the incidence of post-treatment adjacent vertebral fractures have. We have thus developed a method of reconstructing the bolus shape within the bone based on a series of images acquired with fluoroscopic radiography. The cement volume was calculated with a mean error of 8%, and the location and shape of the bolus were visualized in both two and three dimensions. Three-dimensional imaging modalities allow better visualization and diagnostics and can lead to more precise modeling and optimization of surgical procedures. The techniques presented in this thesis aim to make imaging more accessible and broaden the associated range of applications.

The metallofullerenes, Hox C82 (x = 1,2) and Gd C 82 were generated, sublimed from carbon-arc soot, and further purified by HPLC. Sample purity was then judged by LD-TOF MS and compared with purities obtained by ICP-AE and NAA (Ho only) metal assays. Results indicated that the mass spectrometry data overestimated the purity of the metallofullerene samples by about a factor of two. The water-soluble metallofullerol derivatives, Hox C82(OH)y and Gd C82(OH) y were then prepared, and studies were conducted to determine their proton relaxivities in aqueous solution and magnetic properties in the solid state. The relaxivity of Gd C82(OH)y (R1 = 19.9 mM-1 sec.-1 at 20 MHz, 40°C) in aqueous solution was 5 times greater than commercially available MRI contrast agents, indicating that Gd C82(OH)y (6.4 muB at 100K) holds great promise as an MRI contrast agent. In comparison, Hox C82(OH)y did not enhance proton relaxation and had a magnetic moment lower than its Hox C82 precursor (Hox C82(OH)y = 3.8--5.2 muB at 300K, vs. Hox C82 = 5.5--7.8 mu B at 300K). Finally, full biodistribution studies of a water-soluble radioactive metallofullerene compound were conducted using 166Ho x C82(OH)y in BALB/c mice. For this work, the metallofullerol was neutron-activated (165Ho (n,gamma) → 166Ho) to prepare the 166Hox C82(OH) y radiotracer [Egamma(166Ho) = 80.5 keV] and was monitored after intravenous administration for up to 48 hours by dissection radioanalysis for comparison with a Na2[166Ho(DTPA)(H2O)] control study. Results showed 166Ho localization in the liver with slow clearance, as well as slow uptake by bone without clearance. In contrast, excretion of the control compound was nearly quantitative after 1 hour. The fate of 166Ho was also explored by a metabolism study of 166Hox C82(OH)y in Fischer rats. Results demonstrated 20% clearance of the 166Ho within 5 days. Together, these studies comprise a demonstration of metallofullerene materials utility in radiotherapy, radiodiagnostics, and MRI contrast agent applications.