Includes bibliographical references. / Three problems are addressed in this dissertation. They are intracranial volume extraction, noise suppression and automated segmentation of X-Ray Computerized Tomography (CT) images. The segmentation scheme is based on a Seeded Region Growing algorithm. The intracranial volume extraction is based on image symmetry and the noise suppression filter is based on the Gaussian nature of the tissue distribution. Both are essential in achieving good segmentation results. Simulated phantoms and real medical images were used in testing and development of the algorithms. The testing was done over a wide range of noise values, object sizes and mean object grey levels. All the methods were first implemented in two- and then three-dimensions. The 3-D implementation also included an investigation into volume formation and the advantages of 3-D processing. The results of the intracranial extraction showed that 9% of the data in the relevant grey level range consisted of unwanted scalp (The scalp is spatially not part of the intracranial volume, but has the same grey level values). This justified the extraction the intracranial volume for further processing. For phantom objects greater than 741.51mm³ (voxel resolution 0.48mm x 0.48mm x 2mm) and having a mean grey level distance of 10 from any other object, a maximum segmentation volume error of 15% was achieved.

A characteristic feature of Chronic Obstructive Pulmonary Disease (COPD) is an abnormal inflammatory response in the lungs to inhaled particles or gases. The ability to assess and monitor this response in the lungs of COPD patients is important for understanding the pathogenic mechanisms, but also provides a measure of the activity of the disease. Disease activity is more likely to relate to lung inflammation rather than the degree of airflow limitation as measured by the FEV1. Preliminary studies have shown the 18F fluorodeoxyglucose positron emission tomography (18F FDG-PET) signal, as a measure of lung inflammation, is quantifiable in the lungs and is increased in COPD patients compared to controls. However, the methodology requires standardisation and any further enhancement of the methodology would improve its application to assess inflammation in the lungs. I investigated various methods of assessing FDG uptake in the lungs and assessed the reproducibility of these methods, and particularly evaluated whether the data was reproducible or not in the COPD patients (smokers and ex-smokers). This data was then compared with a group of healthy controls to assess the role of dynamic 18F FDG-PET scanning as a surrogate marker of lung inflammation. My data showed a good reproducibility of all methods of assessing FDG lung uptake. However, using conventional Patlak analysis, the uptake was not statistically different between COPD and the control group. Encouraging results in favour of COPD patients were nonetheless shown using compartmental methods of assessing the FDG lung uptake, suggesting the need to correct for the effect of air and blood (tissue fraction effect) when assessing this in a highly vascular organ like the lungs. A prospective study analysis involving a bigger cohort of COPD patients would be desirable to investigate this further.

A reconstruction algorithm in computerized tomography is a procedure for reconstructing the attenuation coefficientscient, a real-valued function associated with the object of interest, from the measured projection data. Generally speaking, reconstruction algorithms in CT fall into two categories: direct, e.g., filtered back-projection (FBP), or iterative. In this thesis, we discuss a new fast matrix-free iterative reconstruction method based on a polyenergetic model. While most modern x-ray CT scanners rely on the well-known filtered back-projection algorithm, the corresponding reconstructions can be corrupted by beam hardening artifacts. These artifacts arise from the unrealistic physical assumption of monoenergetic x-ray beams. In this thesis, to compensate, we use an alternative model that accounts for differential absorption of polyenergetic x-ray photons and discretize it directly. We do not assume any prior knowledge about the physical properties of the scanned object. We study and implement different solvers and nonlinear unconstrained optimization methods, such as a Newton-like method and an extension of the Levenberg-Marquardt-Fletcher algorithm. We explain how we can use the structure of the Radon matrix and the properties of FBP to make our method matrix-free and fast. Finally, we discuss how we regularize our problem by applying different regularization methods, such as Tikhonov and regularization in the 1-norm. We present numerical reconstructions based on the associated nonlinear discrete formulation incorporating various iterative optimization methods.

A reconstruction algorithm in computerized tomography is a procedure for reconstructing the attenuation coefficientscient, a real-valued function associated with the object of interest, from the measured projection data. Generally speaking, reconstruction algorithms in CT fall into two categories: direct, e.g., filtered back-projection (FBP), or iterative. In this thesis, we discuss a new fast matrix-free iterative reconstruction method based on a polyenergetic model. While most modern x-ray CT scanners rely on the well-known filtered back-projection algorithm, the corresponding reconstructions can be corrupted by beam hardening artifacts. These artifacts arise from the unrealistic physical assumption of monoenergetic x-ray beams. In this thesis, to compensate, we use an alternative model that accounts for differential absorption of polyenergetic x-ray photons and discretize it directly. We do not assume any prior knowledge about the physical properties of the scanned object. We study and implement different solvers and nonlinear unconstrained optimization methods, such as a Newton-like method and an extension of the Levenberg-Marquardt-Fletcher algorithm. We explain how we can use the structure of the Radon matrix and the properties of FBP to make our method matrix-free and fast. Finally, we discuss how we regularize our problem by applying different regularization methods, such as Tikhonov and regularization in the 1-norm. We present numerical reconstructions based on the associated nonlinear discrete formulation incorporating various iterative optimization methods.

Prevalence of osteoarthritis (OA) is less common in the ankle compared to other joints; however, deformation brought on by degeneration causes pain, loss of function, and overall decreased quality of life. Current surgical interventions for end-stage ankle OA are not as reliable as surgical treatments for other joints. Ankle arthroplasty currently has high failure rates, and there are lack of substantial data from long-term outcome studies. By understanding the morphometric changes that occur during the different stages of OA, we are able to identify early signs of the disease with the intention to apply treatment earlier in order to preclude the need for end-stage surgical intervention. The goals of this study are to assess morphometric parameters of the talus as it relates to the progression of OA and to evaluate the effect of gender and anatomical side. A retrospective study was performed where data from sixty-eight CT scans were obtained from two study groups, one with OA and one without. The subjects were segmented, standardized, and normalized in order to study several 3D parameters of the talus, including height, radius of curvature, and volume. Results showed that talar morphometry is influenced by gender and that geometric changes are a function of OA progression. The lateral radii of subjects with OA was significantly larger than those of normal ankles (p<0.0001), and there is evidence of inherent changes between KL grades (p=0.0003). Identifying morphometric changes of the talus at each stage of OA can inherently contribute to better understanding the degenerative process. Assessing specific characteristics at earlier stages of the diseases may help clinicians to diagnose more accurately and to better provide treatment.

Midshaft clavicle fractures are very common. Current treatment of choice involves internal fixation with superior or anterior clavicle plating, however their clinical success and patient satisfaction are slowly decreasing. The design of intramedullary (IM) devices is on the rise, but data describing the IM canal parameters is lacking. The aim of this study is to quantify morphometry of the clavicle and its IM canal, and to evaluate the effect of gender and anatomical side. This study used 3-dimensional (3D) image-based models with novel and automated methods of standardization, normalization and bone cross-section evaluation. The data obtained in this thesis presents IM canal and clavicle radius and center deviation parameterized as a function of clavicle length, in addition, its radius of curvature and true length. Results showed that right-sided clavicles tended to be shorter and thicker than left-sided, but only males showed a statistically significant difference in size compared to females (p<.0001). The smallest IM canal and clavicle radii were seen at different clavicle lengths (54% and 49%), suggesting that the narrowest region of IM canal cannot be appreciated based on external visualization of the clavicle alone. The narrowing of the IM canal is of special interest because this a potential limiting region for IM device design. Furthermore, the location and value of maximum lateral curvature displacement is different in the IM canal, implying there exists an eccentricity of the IM canal center with respect to the clavicle center.

COVID-19 is a highly contagious and virulent disease caused by the severe acute respiratory syndrome-coronavirus-2 (SARS-CoV-2). COVID-19 disease induces lung changes observed in lung computerized tomography (CT) and the percentage of those diseased areas on the CT correlates with the severity of the disease. Therefore, segmentation of CT images to delineate the diseased or lesioned areas is a logical first step to quantify disease severity, which will help physicians predict disease prognosis and guide early treatments to deliver more positive patient outcomes. It is crucial to develop an automated analysis of CT images to save their time and efforts. This dissertation proposes CoviNet, a deep three-dimensional convolutional neural network (3D-CNN) to diagnose COVID-19 in CT images. It also proposes CoviNet Enhanced, a hybrid approach with 3D-CNN and support vector machines. It also proposes CoviSegNet and CoviSegNet Enhanced, which are enhanced U-Net models to segment ground-glass opacities and consolidations observed in computerized tomography (CT) images of COVID-19 patients. We trained and tested the proposed approaches using several public datasets of CT images. The experimental results show the proposed methods are highly effective for COVID-19 detection and segmentation and exhibit better accuracy, precision, sensitivity, specificity, F-1 score, Matthew's correlation coefficient (MCC), dice score, and Jaccard index in comparison with recently published studies.