As everyone knows, cancer is one of the greatest health enemies of mankind and became a major public health problem all over the world. Moreover, lung cancer is the most common global cancer leading to more than 1.3 million estimated deaths annually worldwide. Breast cancer is the most common cancer and the 2nd leading cause of death from in women in the US, Canada, and China. Therefore an instrument for the early diagnosis and monitoring of cancers in areas such as lung and breast is immensely important and necessary. The cadmium zinc telluride (CZT) semiconductor is a kind of novel radiation detector that can provide detailed information about spatial position. Also, the energy resulting from interactions between the radiation and the CZT material can also be assessed giving this type of detector added value in disease evaluation. A great deal of work on CZT detectors has been done for breast CT scanning. Furthermore, the CZT detector is also a potential solution for problems in lung cancer CT evaluation where the lack of energy information and high radiation exposure are less than ideal. Since sub-millimetre spatial resolution is required for post-contrast thoracic CT assessment, and $100~\mu m$ or better is critical for breast CT, the investigation of spatial resolution for CZT detectors is essential. In this thesis, the interactions within CZT detectors were analyzed. The influence of different configurations, including adjustment of anode size, gap size, detector thickness and bias voltage, were discussed. Next the methods to evaluate two kinds of signals (collected and transit signals), are provided in Chapter~2. In Chapter~3, an intensity difference method and an intensity ratio method for estimating the sub-pitch spatial resolution were described in detail. Finally, a detector spatial resolution between $10~\mu m$ and $20~\mu m$ was achieved using the collecting signal intensities ratio method and a $4~mm$ thick CZT detector with gap size of $g=50~\mu m$ and bias voltage of $V=300V$. Future work should focus on the contributions from characteristic X-rays emitted from cadmium and tellurium atoms. Also, the work presented was only on pixelated arrays and further assessment of cross-strip electrode detectors could also be of benefit. Lastly based on simulations done in this thesis all photons were assumed to strike the cathode at 90 degrees. Future work should also include non-orthogonal directions for photons. / Thesis / Master of Applied Science (MASc)

Master of Science / Department of Biological & Agricultural Engineering / Kyle R. Douglas-Mankin / Radar-derived P3 data from Next Generation Radar (NEXRAD) of the National Weather Service (NWS) offer higher spatial resolution than precipitation gauge data, which might improve the accuracy of streamflow simulations using watershed models. The objective of this study was to evaluate the performance of spatially-averaged subwatershed-specific NEXRAD P3 data on streamflow simulations using Soil and Water Assessment Tool (SWAT) model. The SWAT hydrologic model was chosen for this study to simulate the hydrologic processes in North Fork Ninnescah Watershed located in south-central Kansas. A precipitation gauge station for each subwatershed was created using an area-weighted average of NEXRAD P3 precipitation estimates for all HRAP grid cells covering the subwatershed. The SWAT model was calibrated with both NEXRAD P3 data and NCDC precipitation gauge (PG) data from 1 January 2002 to 31 December 2008. The P3-calibrated model was validated using PG data for the same simulation period (2002-2008), and vice versa. The PG-calibrated model yielded slightly higher daily Nash-Sutcliffe efficiency (E(subscript)NS = 0.40) than P3 calibrated model (ENS = 0.35), but the yearly ENS and PBIAS for P3 calibrated model (ENS = 0.80) was much better than PG-calibrated model (ENS = 0.43). The P3-validated model (PG calibration) had yearly ENS = of 0.70, whereas the PGcalibrated model had ENS = 0.43. The daily PBIAS value for P3-calibrated model in 2007 (wet year) was -14.13 and for the P3-calibrated model was -32.83; PG data overestimated the streamflow compared to P3 data in 2007. The P3 data has better agreement with PG data from 2002-2008 period than for 1996-2001 period. The streamflow estimation was better with NEXRAD P3 precipitation data in both calibration and validation runs. Even though the model was calibrated with PG data, the validated model with P3 data has comparatively high ENS. The spatial variation of streamflow response within the watershed was greater compared to the temporal variation in both the calibrated models. The spatial representation of precipitation data by NEXRAD P3 has improved the modeling performance compared to PG data; it is evident that NEXRAD data is an alternative to precipitation gauge measurements.

Seismic survey design methods often rely on qualitative measures to provide an optimal image of their objective target. Fold, ray tracing techniques counting ray hits on binned interfaces, and even advanced 3-D survey design methods that try to optimize o?set and azimuth coverage are prone to fail (especially in complex geological or structural settings) in their imaging predictions. The reason for the potential failure of these commonly used approaches derives from the fact that they do not take into account the ray geometry at the target points. Inverse theory results can provide quantitative and objective constraints on acquisition design. Beylkin??s contribution to this ?eld is an elegant and simple equation describing a reconstructed point scatterer given the source/receiver distribution used in the imaging experiment. Quantitative measures of spatial image resolution were developed to assess the e?cacy of competing acquisition geometries. Apart from the source/receiver con?guration, parameters such as the structure and seismic velocity also in?uence image resolution. Understanding their e?ect on image quality, allows us to better interpret the resolution results for the surveys under examination. A salt model was used to simulate imaging of target points located underneath and near the ?anks of the diapir. Three di?erent survey designs were examined. Results from these simulations show that contrary to simple models, near-o?sets do not always produce better resolved images than far-o?sets. However, consideration of decreasing signal-to-noise ratio revealed that images obtained from the far-o?set experiment are degrading faster than the near-o?set ones. The image analysis was performed on VSP ?eld data as well as synthetics generated by ?nite di?erence forward modeling. The predicted image resolution results were compared to measured resolution from the migrated sections of both the ?eld data and the synthetics. This comparison con?rms that image resolution analysis provides as good a resolution prediction as the prestack Kirchho? depth migrated section of the synthetic gathers. Even in the case of the migrated ?eld data, despite the presence of error introducing factors (di?erent signal-to-noise ratios, shape and frequency content of source wavelets, etc.), image resolution performed well exhibiting the same trends of resolution changes at di?erent test points.

Seismic survey design methods often rely on qualitative measures to provide an optimal image of their objective target. Fold, ray tracing techniques counting ray hits on binned interfaces, and even advanced 3-D survey design methods that try to optimize o?set and azimuth coverage are prone to fail (especially in complex geological or structural settings) in their imaging predictions. The reason for the potential failure of these commonly used approaches derives from the fact that they do not take into account the ray geometry at the target points. Inverse theory results can provide quantitative and objective constraints on acquisition design. Beylkin??s contribution to this ?eld is an elegant and simple equation describing a reconstructed point scatterer given the source/receiver distribution used in the imaging experiment. Quantitative measures of spatial image resolution were developed to assess the e?cacy of competing acquisition geometries. Apart from the source/receiver con?guration, parameters such as the structure and seismic velocity also in?uence image resolution. Understanding their e?ect on image quality, allows us to better interpret the resolution results for the surveys under examination. A salt model was used to simulate imaging of target points located underneath and near the ?anks of the diapir. Three di?erent survey designs were examined. Results from these simulations show that contrary to simple models, near-o?sets do not always produce better resolved images than far-o?sets. However, consideration of decreasing signal-to-noise ratio revealed that images obtained from the far-o?set experiment are degrading faster than the near-o?set ones. The image analysis was performed on VSP ?eld data as well as synthetics generated by ?nite di?erence forward modeling. The predicted image resolution results were compared to measured resolution from the migrated sections of both the ?eld data and the synthetics. This comparison con?rms that image resolution analysis provides as good a resolution prediction as the prestack Kirchho? depth migrated section of the synthetic gathers. Even in the case of the migrated ?eld data, despite the presence of error introducing factors (di?erent signal-to-noise ratios, shape and frequency content of source wavelets, etc.), image resolution performed well exhibiting the same trends of resolution changes at di?erent test points.

For several common inkjet, laser and thermal dye printers, a method of evaluating prints that is not associated with the level of dental expertise of the observer is introduced. In addition, an automated analysis that mimics the observations made by observers is tested. Metrics that are evaluated in this study include spatial resolution, contrast uniformity, the type of paper, and overall observer preference. The results demonstrate that observer preference is associated with a high print contrast uniformity and with the use of glossy paper, but not with increased spatial resolution. The automated analysis produced results that were in general agreement with the observer data for spatial resolution, which concluded that the Lexmark C543DN printer produced prints with the highest spatial resolution. A thermal dye printer (Kodak CMI1000) produced prints with the highest contrast uniformity, and the print most favored by observers overall was produced by the Kodak ESP-9 inkjet printer on Kodak Everyday Glossy Photo paper.

For several common inkjet, laser and thermal dye printers, a method of evaluating prints that is not associated with the level of dental expertise of the observer is introduced. In addition, an automated analysis that mimics the observations made by observers is tested. Metrics that are evaluated in this study include spatial resolution, contrast uniformity, the type of paper, and overall observer preference. The results demonstrate that observer preference is associated with a high print contrast uniformity and with the use of glossy paper, but not with increased spatial resolution. The automated analysis produced results that were in general agreement with the observer data for spatial resolution, which concluded that the Lexmark C543DN printer produced prints with the highest spatial resolution. A thermal dye printer (Kodak CMI1000) produced prints with the highest contrast uniformity, and the print most favored by observers overall was produced by the Kodak ESP-9 inkjet printer on Kodak Everyday Glossy Photo paper.

Imaging of the osseous structures of the temporomandibular joint is best accomplished using computed tomography (CT). Cone beam CT offers a reduced radiation dose and improved spatial resolution compared to multislice helical CT. This study evaluates comparative dosimetry for temporomandibular joint imaging using two different cone beam CT systems, the Hitachi CB MercuRay and Kodak 9000 3D. These systems demonstrate differing properties with respect to field-of view sizes, operational technique factors, and spatial resolution. The Kodak 9000 3D unit offers an effective radiation dose reduction of greater than ten-fold compared with the Hitachi CB MercuRay, depending on kVp and mA. A subsequent clinical study evaluating the effect of spatial resolution on the ability to detect osseous changes related to temporomandibular joint degenerative disease found no significant difference in diagnostic efficacy between high and low spatial resolution images, however, observers consistently associated high spatial resolution with superior image quality.

Imaging of the osseous structures of the temporomandibular joint is best accomplished using computed tomography (CT). Cone beam CT offers a reduced radiation dose and improved spatial resolution compared to multislice helical CT. This study evaluates comparative dosimetry for temporomandibular joint imaging using two different cone beam CT systems, the Hitachi CB MercuRay and Kodak 9000 3D. These systems demonstrate differing properties with respect to field-of view sizes, operational technique factors, and spatial resolution. The Kodak 9000 3D unit offers an effective radiation dose reduction of greater than ten-fold compared with the Hitachi CB MercuRay, depending on kVp and mA. A subsequent clinical study evaluating the effect of spatial resolution on the ability to detect osseous changes related to temporomandibular joint degenerative disease found no significant difference in diagnostic efficacy between high and low spatial resolution images, however, observers consistently associated high spatial resolution with superior image quality.

De nombreuses applications pratiques ou industrielles, telles que l’étude de la dispersion atmosphérique de polluants, la qualité de l’air, la micro-météorologie en terrain complexe et l’évaluation du potentiel éolien, nécessitent la prédiction précise de l’écoulement atmosphérique à une échelle dite locale (environ 10 km horizontalement). Le travail de recherche dans le cadre de cette thèse s’inscrit donc dans la proposition d’une chaine de méthodologies et de modélisations permettant de simuler l’écoulement atmosphérique à cette échelle, avec une résolution spatiale horizontale hectométrique. Tout d’abord, nous nous sommes intéressés à la modélisation de la turbulence dans la couche limite atmosphérique (CLA). Pour cela, nous avons choisi le modèle RANS k− ε (déjà largement utilisé dans la littérature), ainsi que le modèle RANS Ri j − ε afin de simuler l’anisotropie de la turbulence. Nous avons ainsi pu vérifier la nécessité d’utiliser les constantes de Duynkerke (1988) pour l’atteinte des niveaux de turbulence atmosphérique avec le modèle k− ε. Dans cette optique, nous avons également développé un nouveau jeu de constantes atmosphériques pour le modèle Ri j − ε. Finalement, nous avons proposé un modèle théorique capable de reproduire les caractéristiques turbulentes de l’écoulement pour n’importe quel temps d’intégration, permettant ainsi de trouver une continuité entre les constantes « standards » et les constantes « atmosphériques » des modèles de turbulence. D’autre part, nous avons développé l’approche de modélisation « CFD 1D-3D », qui consiste en l’utilisation d’un modèle CFD 1D afin de fournir les profils verticaux nécessaires pour forcer le code CFD 3D en données météorologiques (utilisé en topographie complexe). Le modèle 1D a été développé au cours de cette thèse avec les modèles de turbulence k− ε et Ri j− ε. Il a été validé grâce à une comparaison avec des résultats empiriques et théoriques issus de la littérature. Cette comparaison a montré des résultats très encourageants de ce modèle dans la simulation de la CLA en sol plat. De plus, la méthodologie « CFD 1D-3D » a été évaluée grâce à une comparaison avec des mesures en soufflerie en présence d’un relief complexe : les résultats sont globalement très satisfaisants. Ces comparaisons ont permis enfin de valider le nouveau jeu de constantes pour le modèle Ri j− ε. Finalement, nous nous sommes intéressés à l’utilisation de calculs CFD partiellement convergés comme moyen de réduction du temps CPU des codes CFD, dans des contextes d’utilisation opérationnelle. Dans cette optique, nous avons montré que l’on arrive à une solution dont l’erreur est faible par rapport à la solution convergée (< 10% d’erreur), avec un temps CPU de l’ordre de 5%−10% du temps nécessaire pour atteindre la convergence. C’est un résultat très intéressant car il permet de réduire considérablement le temps de calcul, tout en gardant une erreur faible devant l’incertitude générale de l’approche CFD. / Many practical and industrial applications, such as the study of atmospheric dispersion of pollutants, air quality,micro-meteorology in complex terrain and wind assessment, require accurate prediction of the atmospheric flow at a so-called local scale (approximately 10 km horizontally). Therefore, the main objective in this thesis is to propose a chain of methodologies capable of simulating the atmospheric flow at this scale, with a horizontal hectometric spatial resolution. First of all, we were interested in modeling of turbulence in the atmospheric boundary layer (ABL). In addition to the largely used RANS k−ε model, we considered the use of the RANS Ri j− ε model as a way of simulating turbulence anisotropy.We were able to verify the necessity of using the Duynkerke (1988) constants in order to achieve atmospheric levels of turbulence with the k− ε model. In a similar way, we also developed a new set of atmospheric constants for the Ri j− ε model. Finally, we proposed a theoretical model capable of reproducing the main characteristics of a turbulent flow for any given sampling duration, thus allowing a more continuous approach between « standard » and « atmospheric » constants for turbulence models. Also, in this thesis, we developed the « CFD 1D-3D » modeling approach. It is based on the use of a 1D CFD model as a way of providing vertical profiles of meteorological data for boundary conditions of a 3D CFD code, used in complex terrain. This 1D model was developed as a part of the thesis, along with k− ε and Ri j − ε turbulence models. It was validated by being compared with empirical and theoretical results. The comparisons showed very encouraging results concerning the ability of this model in simulating ABL in the presence of a flat terrain. In addition, the « CFD 1D- 3D » methodology was assessed by comparison with wind tunnel measurements in the presence of complex terrain, which showed very satisfactory resultst. These comparisons also validated the newly developed set of constants for the Ri j− ε model. Finally, we studied the use of partially converged CFD as a way of reducing the CPU time of CFD simulations for operational purposes. We therefore demonstrated that we can achieve a low error solution (< 10% error compared with the converged solution), with a CPU time of about 5%−10% of the time required to achieve convergence. This result was very interesting because the methodology significantly reduces the computational time while maintaining a low error as compared to the overall uncertainty of the CFD approach.

No description available.