Search

Brain metastases (BMs) represent advanced cancer, and so BM patients must be treated quickly and effectively while minimizing treatment toxicities. Stereotactic radiosurgery (SRS) uses conformal, ablative radiation doses to treat BMs, but with failure rates up to 30%. A predictive model of BM progression post-SRS would therefore aid in BM treatment selection and SRS planning. Previous studies have used pre-treatment T1-weighted contrast-enhanced magnetic resonance imaging (T1w-CE MRI) to predict SRS outcomes, through quantitative radiomic analysis with machine learning (ML) and qualitative appearance analysis. Comparison of these methods is not well understood, and ML methods have not been studied for sensitivity to relevant clinical factors, robust model interpretability, or multi-centre external validation. To meet these needs, a dataset of 123 BMs across 99 SRS patients was used to develop T1w-CE MRI radiomics-based ML models to understand their sensitivity to clinical factors. A ML model using radiomic and clinical features obtained the highest area-under-the-receiver-operating-characteristic-curve (AUC) of 0.77, and this model was sensitive to primary cancer site, BM volume, and MRI scanner model. BM volume sensitivity was reduced by removing volume-correlated radiomic features. An observer study of BM qualitative analysis revealed high interobserver variability that in turn limited outcome prediction, while ML models provided enhanced stratification of BMs into risk groups for post-SRS progression (Kaplan-Meier rank-sum p = 0.0003). BM qualitative appearance was useful for ML model interpretation, revealing the post-SRS progression radiomic signature is tied to necrotic or heterogeneous BM appearance, indicating a potentially less radiosensitive, hypoxic environment. An additional dataset of 117 BMs across 62 SRS patients was collected at a second centre for external validation. Transferring a locked model between centres revealed poor performance, but limiting the model to use radiomic features important at both centres increased the AUC to 0.70. Retraining a model with the second centre’s dataset using a locked methodology developed with the first centre’s dataset achieved a higher AUC of 0.80. In conclusion, this work successfully characterized radiomics-based ML model performance with respect to clinical factors and BM qualitative appearance, while also providing the ML model interpretability and external validation necessary to motivate future research and clinical translation.

Background: Skeletal muscle (SM), with its precise O2 supply-demand matching, is ideal for studying microvascular (MV) function, which is crucial in cardiovascular physiology and disease. Our goal is to gain understanding of SM microcirculation using recent capillary network module (CM) data from rat in a computational model. Methods: We construct a 4-CM network, with single-vessel equivalents for each CM, three arterioles, and two venules. A two-phase (plasma/RBCs) steady-state model is used to calculate blood flow. An iterative boundary pressure finding method is developed to match flow in CMs. Results: We validate our flow and pressure models vs. experimental data, and show how inflow hematocrit affects resistance and RBC distribution. We show that venular pressure regulation is needed to control individual CM RBC flow. Discussion: Our computational model sheds new light on flow and regulation in interconnected CMs, and supports future studies using more CMs or time-dependent flow and regulation.

Neuromonitoring during surgery is used to detect early indications of cerebral injury before permanent damage occurs. A commonly used technology is cerebral oximetry; however, current systems only monitor one brain region and have limited depth sensitivity. A newly developed NIRS system, Kernel Flow, offers the possibility to address both limitations by providing full-head coverage and time-resolved detection to enhance sensitivity to the brain. This work aimed to assess Kernel Flow’s sensitivity to regional cerebral oxygenation changes. Two experiments were conducted. In the first, decreases in cerebral oxygenation caused by transient carotid compression were measured in healthy volunteers. The second was a clinical feasibility study in which the device was used to detect possible regional differences in cerebral desaturation in patients undergoing shoulder surgery. Overall, Kernel Flow showed good sensitivity to regional changes in cerebral oxygenation – although hair provided a practical challenge – and has promise as an intraoperative neuromonitor.

Fetal life is a significant period of human development because organ systems that sustain life outside the uterus begin to develop during this time. A necessary process that begins during fetal life is myelination, which is the process by which myelin, a lipid-rich substance, is wrapped around neurons in the brain to increase the speed of action potential transmission. Since myelination is critical for the normal function of the central nervous system, fetal myelin assessment is important for understanding neurodevelopment and neurodegeneration, such as intrauterine growth restriction (IUGR). Magnetic resonance imaging (MRI) is an excellent tool for visualizing fetal anatomy and identifying pathology. Relaxometry quantifies T1, T2, and T2* relaxation times, which are MR parameters that reflect fundamental tissue properties sensitive to the tissue microenvironment, providing an interpretation of images in absolute units. For the first time, fetal tissue T1 and T2* relaxation times were successfully quantified in uncomplicated pregnancies as a function of gestational age (GA) in the third trimester. A tissue microenvironment that can be investigated by MR relaxometry is myelin water. Myelin water imaging (MWI) uses MR relaxometry to visualize the aqueous components associated with myelin sheath to quantify myelin water fraction (MWF), a validated myelin marker. Moving to a guinea pig model of pregnancy, MWI was successfully conducted in the fetal environment as MWF was quantified in various fetal brain regions late in gestation. To investigate the effects of IUGR on myelination in utero, MWI was applied in a guinea pig model of natural IUGR late in gestation. MWF was significantly reduced in different brain regions of guinea pigs with IUGR compared to those without IUGR. Furthermore, the study highlighted the utility of MWF as a functional marker for IUGR. In conclusion, this dissertation demonstrates using MR relaxometry to quantify T1 and T2* relaxation times of fetal tissues throughout pregnancy and assess fetal brain myelin content in both a normal and IUGR environment. The findings demonstrate MR relaxometry's utility in assessing fetal tissue development in utero.

The successful intervention of breast cancer relies on effective early detection and definitive diagnosis. While conventional screening mammography has substantially reduced breast cancer-related mortalities, substantial challenges persist in women with dense breasts. Additionally, complex interrelated risk factors and healthcare disparities contribute to breast cancer-related inequities, which restrict accessibility, impose cost constraints, and reduce inclusivity to high-quality healthcare. These limitations predominantly stem from the inadequate sensitivity and clinical utility of currently available approaches in increased-risk populations, including those with dense breasts, underserved and vulnerable populations. This PhD dissertation aims to describe the development and validation of alternative, cost-effective, robust, and high-resolution systems for point-of-care (POC) breast cancer screening and image-guided needle interventions. Specifically, 2D and 3D ultrasound (US) and positron emission mammography (PEM) were employed to improve detection, independent of breast density, in conjunction with mechatronic and automated approaches for accurate image acquisition and precise interventional workflow. First, a mechatronic guidance system for US-guided biopsy under high-resolution PEM localization was developed to improve spatial sampling of early-stage breast cancers. Validation and phantom studies showed accurate needle positioning and 3D spatial sampling under simulated PEM localization. Subsequently, a whole-breast spatially-tracked 3DUS system for point-of-care screening was developed, optimized, and validated within a clinically-relevant workspace and healthy volunteer studies. To improve robust image acquisition and adaptability to diverse patient populations, an alternative, cost-effective, portable, and patient-dedicated 3D automated breast (AB) US system for point-of-care screening was developed. Validation showed accurate geometric reconstruction, feasible clinical workflow, and proof-of-concept utility across healthy volunteers and acquisition conditions. Lastly, an orthogonal acquisition and 3D complementary breast (CB) US generation approach were described and experimentally validated to improve spatial resolution uniformity by recovering poor out-of-plane resolution. These systems developed and described throughout this dissertation show promise as alternative, cost-effective, robust, and high-resolution approaches for improving early detection and definitive diagnosis. Consequently, these contributions may advance breast cancer-related equities and improve outcomes in increased-risk populations and limited-resource settings.

Cell therapies, including mesenchymal stem cells (MSC) and regulatory T cells (Tregs) have shown potent and long-lasting therapeutic benefit in several disease applications. Unfortunately, the effectiveness of these therapies is variable and following administration, it is largely unknown where therapeutic cells traffic and how many persist over time. This thesis aims to advance and compare iron- and fluorine-19 (19F)-based magnetic resonance imaging (MRI) and magnetic particle imaging (MPI) for therapeutic cell tracking. First, we developed a trimodal imaging approach to study MSC fate in vivo. In a mouse model, iron-labeled MSC were detected by MRI and MPI, and each modality provided unique information on MSC quantity (MPI) and anatomical localization (MPI) in vivo. In these same mice, 19F MRI was used to assess inflammatory influx to the injection site after intravenous administration of 19F-perfluorocarbon agent. Next, the sensitivity and cellular detection limits of 19F MRI were compared to MPI. MPI showed superior sensitivity, as fewer iron-labeled cells could be detected with MPI (4000 MSCs) compared to 19F-labeled cells with 19F MRI (256,000 MSCs) using the same imaging time. For cell tracking, estimation of cell number is critical, so we sought to develop and test four methods to select and quantify MPI signal from an image. In this project, it is demonstrated that while MPI signal scales directly linear with the amount of iron present, 3 users showed more accurate and precise quantification when they considered a larger area for each signal in the image. In the final chapter, we demonstrate a clinically-applicable 19F MRI imaging approach for tracking in vivo biodistribution of Tregs. This approach involved a dual tuned MRI surface coil with sensitive bSSFP imaging sequence on a 3 Tesla human MRI. The ability to non-invasively quantify and image injected cells in vivo with high sensitivity will improve the efficacy and safety of cell therapy.

The Foster lab has been developing cellular MRI tools for well over a decade using superparamagnetic iron oxide (SPIO)-based contrast agents for numerous applications cell tracking applications. SPIO are used to label cells which are indirectly detected as regions of signal loss. The main challenge with this technique is quantification of cell number, which is important for applications such as dendritic cell (DC)-based immunotherapies where cell number is used as a predictor of immunotherapeutic response. Magnetic particle imaging (MPI) is a novel pre-clinical modality that shows promise for cell tracking. MPI directly detects the nonlinear magnetization response of superparamagnetic iron oxide nanoparticles (SPIO), without background signal or attenuation from tissues. This allows for high cellular sensitivity, specificity, and quantification of iron. With these advantages, MPI overcomes the challenges of cellular magnetic resonance imaging (MRI). This thesis describes our three goals for implementing MPI as a sensitive and quantitative technique for cell tracking: (i) test commercially available SPIO for MPI, (ii) quantify DC migration in vivo and (iii) test and validate novel image parameters. We contributed to the advancement of this technology by detecting as few as 1000 DC (4.4 ng) in vivo. With clinical MPI systems currently being developed, this work contributed significantly towards the development, validation, and implementation of MPI as an advanced therapeutic cell tracking modality.

The use of anti-cancer non-ablative electric fields is an expanding area of research that includes clinically available external devices for the treatment of glioblastoma (GBM), and a pre-clinical internal system called Intratumoural Modulation Therapy (IMT). IMT uses multiple electrodes implanted within the tumour to apply low intensity electric fields (~1 V/cm) focused on the target region, anywhere in the brain, with no externally visible devices. In this thesis, multi-electrode spatiotemporally dynamic IMT is investigated through computer simulation, numerical optimization, brain phantom and in vitro validation methods. These planning and validation strategies are hypothesized to improve tumour coverage with the necessary electric field and improving treatment efficiency through minimizing number of electrodes, power consumption, and manual planning time. The development of an IMT optimization algorithm that considers the placement of multiple electrodes, voltage amplitude and phase shift of input waveforms showed that human scale tumours are coverable with anti-cancer electric fields. Additionally, maximally separating the relative phase shifts of sinusoidal voltage waveforms applied to the electrodes, induces rotating electric fields that cover the tumour over time, with spatially homogeneous time averaged fields. A treatment planning system designed specifically for IMT considered optimizable electrode trajectories and patient images to create custom field plans for each patient, which was validated using robotic electrode implantation on a brain phantom. These custom fields can be optimized to conform to patient-specific tumour size, shape, or location. The efficacy of spatiotemporally dynamic fields was validated by developing a purpose-built in vitro device to deliver multi-electrode IMT to patient derived GBM cells. Cell viability was reduced when subjected to these rotating electric fields, supporting the optimality criteria derived analytically. The IMT optimization algorithm and planning system, supporting phantom validation and in vitro data, together with an accompanying planning system user guide support the move to clinical trials in the future. Overall, IMT technology has been advanced in this thesis to include patient-specific treatment planning optimization, a development that holds significance towards the future clinical implementation of IMT and treatment goals.

Introduction: Molecular imaging allows for non-invasive longitudinal visualization of cellular functions in vivo. This area of research has provided better understanding of fundamental molecular and biochemical processes, enabled optimization of pre-clinical and clinical assessments for new treatments, and allowed for more accurate and early detection of many pathologies. Extensive research for novel imaging techniques and emerging technologies have rapidly advanced the field. However, an ideal single imaging modality or approach does not exist. Alternatively, multi-modal imaging approaches are commonly applied to overcome limitations of individual technologies. In this thesis, we design, develop, validate, and image various cell types using multi-modal complementary tracking systems including bioluminescence imaging (BLI), magnetic resonance imaging (MRI), magnetic particle imaging (MPI), and positron emission tomography (PET) to maximize the strengths of each technology. Methods and Results: First, we develop a multi-modal imaging approach to track breast cancer cells in a safe, sensitive, and quantitative manner in vivo. We have subsequently optimized and applied our multi-modal system further to visualize mesenchymal stem cells (MSCs), a clinically-relevant cell type, with high sensitivity. Furthermore, we have validated this imaging approach in an emerging cellular immunotherapy for the treatment of a rodent model of ovarian cancer. Specifically, using a safe dual human reporter-gene imaging approach, breast cancer cells were detectable in mice with PET and MRI - two clinical imaging modalities. Next, MSCs were genetically modified with a PET reporter gene and labelled with iron oxide nanoparticles for longitudinal imaging with clinical PET and sensitive MPI. Finally, we demonstrate chimeric antigen receptor natural killer (CAR-NK) cells slowed tumor progression in an ovarian cancer model using BLI, and were able to track the CAR-NK cells using reporter-based BLI and PET. Conclusion: The studies reported in this thesis contribute new multi-modal non-invasive molecular imaging tools to track various cell types in preclinical models to reveal complementary information on cell localization, proliferation, viability, and therapeutic response. Continued development of the clinically-relevant imaging tools we have built for tracking MSCs and CAR-NK cells may one day provide valuable information on cell therapy response/non-response or side effects in individual patients – important goals in the era of precision medicine.

Introduction: Breast cancer remains a leading cause of mortality among women due to the propensity of breast tumours to spread and metastasize to distant sites, including the brain. Early detection of metastatic disease has been challenging, and typical methods of treatment often fail. A significant roadblock in advancing the detection and treatment of breast cancer brain metastases is the lack of representative experimental preclinical models and methods of studying its progression in vivo. Methods: First, we use iron-based cellular MRI to noninvasively track the progression of a brain metastatic breast cancer cell line in murine models with varying immune competencies. We then employ longitudinal BLI and MPI to visualize the tumour progression of a patient-derived xenograft (PDX) model of breast cancer brain metastasis. Finally, we used MPI to evaluate the magnetic performance of iron oxide particles following cell labeling with a brain metastatic breast cancer cell line to optimize cellular detection and cell tracking techniques. Results: Cellular MRI revealed significant differences in tumour progression throughout the brain and body between murine models. We then developed a novel method for labeling PDX cells with iron oxide particles and use MPI and BLI to provide measures of iron content and cellular viability. Finally, we demonstrate that cell labeling can change the magnetic performance of iron agents used for cell tracking with MPI. Conclusion: Comprehensive analysis of cancer cell arrest, clearance, and tumour progression with cellular imaging is important for understanding the metastatic cascade of a model of breast cancer brain metastasis in both cell lines and PDX models. Furthermore, we provide evidence that iron oxide particles are valuable tracers for MPI cell tracking, but their MPI performance may be altered following cell labeling.

Patients diagnosed with non-spine bone metastases benefit from stereotactic body radiotherapy (SBRT), however, 5-20% of patients still develop local progression. We aim to create a machine learning model integrating clinical, radiotherapy dose, and radiomic features to predict which patients will experience local progression following SBRT. We analyzed 179 treated metastatic lesions in 130 patients. Radiomic features were extracted from radiotherapy target volumes. Lesions were separated by patient into an approximate 70/30 training (n = 119) and testing (n = 60) split. Feature selection was performed, and machine learning models were created using dose, clinical, and radiomics features. The clinical model achieved a testing AUC of 0.61 [95% CI: 0.47 – 0.74] and the combined model had a testing AUC of 0.77 [95% CI: 0.64 – 0.89]. This preliminary study demonstrates the potential utility for a radiomics-based machine learning model for the prediction of local progression in non-spine bone metastases following SBRT.

Total knee arthroplasty (TKA) is the only solution for treating arthritis of the knee joint. Although it is successful at reducing pain and returning function to affected joints, one in five patients still report dissatisfaction following their operation. Bicruciate stabilized (BCS) TKA was developed to improve outcomes by replicating normal knee structure and function. The biomechanical, functional, and clinical outcomes for the BCS design were investigated in this thesis through radiographic imaging techniques, wearable sensor systems, and questionnaires in a cohort of TKA patients. A stereo x-ray technique, called radiostereometric analysis (RSA), assesses implant fixation by tracking micromotion of TKA devices relative to the bone. Risk of implant loosening can be predicted based on the magnitude of these micromotions. This thesis found micromotion of the BCS TKA was within safe thresholds for both the gap balancing and measured resection techniques, indicating sufficient fixation to the bone occurs and the BCS TKA is not expected to have elevated revision risks due to implant loosening. The exact cause of patient dissatisfaction after TKA is unknown. This thesis sought to find any differences in objective data between satisfied and dissatisfied patients with a BCS TKA. RSA was used to measure implant micromotion and tibiofemoral contact kinematics. A sensor system tracked measures of patient function during a timed-up-and-go functional test, and patient-reported outcomes were collected. We found no difference in implant micromotions or patient function between satisfied and dissatisfied patients. However, dissatisfied patients had more anterior contact on the lateral condyle of the knee in early flexion, and more pain and unmet expectations. Finally, correlations were found between implant micromotion and tibiofemoral contact kinematics. Contact patterns indicating reduced posterior femoral rollback in the lateral compartment correlated with greater implant micromotion. Since BCS TKA aims to replicate normal knee kinematics and guide posterior rollback, it was concluded that undesired kinematics resulted in greater micromotions, and a greater risk of implant loosening. Overall, the restoration of kinematics—particularly in the lateral compartment—in BCS TKA appears to be important for reducing implant migrations, improving pain and feeling in the knee, and ultimately, enhancing patient satisfaction.

With its superb spatial and temporal resolution, magnetic resonance imaging (MRI) has great potential to track cellular activities that define early stages of disease. To improve molecular imaging techniques, we are developing MRI reporter gene expression based on the magnetosome. In magnetotactic bacteria (MTB), magnetosome formation compartmentalizes iron biominerals in membrane-enclosed vesicles. We hypothesize that essential magnetosome proteins interact in any cell type to form rudimentary magnetosome-like nanoparticles, providing a genetically-controlled contrast agent for molecular MRI. MTB genes mamE, mamB, mamI, and mamL were cloned from M. magneticum sp. AMB-1 genomic DNA by PCR and inserted into fluorescent vectors to create Mam fusion proteins then stably expressed in human MDA-MB-435 melanoma cells. Cines of fluorescent elements detected in intact cells were captured with confocal microscopy (Nikon A1R) and analyzed using both ImageJ and Mathematica for Brownian motion and velocity. To obtain longitudinal and transverse relaxation rates, cells stably expressing magnetosome proteins were supplemented with 250 µM ferric nitrate, harvested, mounted in a gelatin phantom, and scanned at 3 Tesla (Biograph mMR). Tomato-MamL, Tomato-MamL/GFP-MamI, and Tomato-MamB all express punctate, mobile fluorescence, while GFP-MamE expresses punctate but stationary fluorescence. Analysis of motility revealed that magnetosome proteins have variable diffusion coefficients due to their variable sizes, but all magnetosome proteins travel at a velocity of around 0.2 µm/s. Relaxation rates of iron-supplemented cells expressing Tomato-MamB, GFP-MamI, or Tomato-MamL have significantly higher R2 and R2* than non-supplemented cells. Interestingly, iron-supplemented cells expressing GFP-MamE or co-expressing FLAG-MamL/GFP-MamI had relaxation rates comparable to unsupplemented cell types. This is the first report characterizing essential magnetosome proteins MamE, MamB, MamI, and MamL in mammalian cells. Analysis of motion shows that magnetosome proteins travel at velocities comparable to the mammalian motor protein myosin. Expression of either MamB, MamI or MamL increases transverse relaxation rates; however, co-expression of MamI and MamL reduces them again, suggesting a regulatory effect of magnetosome gene combinations. Biosynthesis of magnetosome-like nanoparticles in mammalian cells would provide an endogenous magnetic resonance (MR) contrast agent under genetic control. This patented technology would provide long-term molecular imaging for tracking cellular and molecular activities throughout the cell's life cycle.

Osteoarthritis (OA) is a joint disease that causes pain, stiffness, and loss of function. Inflammation of the synovium plays a role in the pathology of OA. Macrophages are the dominant immune cells in synovial tissue. Activated macrophages over-express the translocator protein (TSPO). [18F]FEPPA is a 2nd generation positron emission tomography (PET) tracer that can target TSPO with high specificity. Hybrid [18F]FEPPA PET/MRI may enable accurate quantification of macrophage activity in vivo. In this work, [18F]FEPPA tracer uptake in knee synovial tissue was measured ex vivo using autoradiography and was validated to correlate to the true macrophage activity determined by immunofluorescence (IF). A clinical [18F]FEPPA PET/MRI protocol showed similar ability to quantify macrophage activation in vivo. This study suggests that [18F]FEPPA PET/MRI may be an effective tool to investigate the role of inflammation in the pathology of OA or to validate the efficacy of future anti-inflammatory OA treatments.

Gas-exchange is the primary function of the lungs and involves removing carbon dioxide from the body and exchanging it within the alveoli for inhaled oxygen. Several different pulmonary, cardiac and cardiovascular abnormalities have negative effects on pulmonary gas-exchange. Unfortunately, clinical tests do not always pinpoint the problem; sensitive and specific measurements are needed to probe the individual components participating in gas-exchange for a better understanding of pathophysiology, disease progression and response to therapy. In vivo Xenon-129 gas-exchange magnetic resonance imaging (129Xe gas-exchange MRI) has the potential to overcome these challenges. When participants inhale hyperpolarized 129Xe gas, it has different MR spectral properties as a gas, as it diffuses through the alveolar membrane and as it binds to red-blood-cells. 129Xe MR spectroscopy and imaging provides a way to tease out the different anatomic components of gas-exchange simultaneously and provides spatial information about where abnormalities may occur. In this thesis, I developed and applied 129Xe MR spectroscopy and imaging to measure gas-exchange in the lungs alongside other clinical and imaging measurements. I measured 129Xe gas-exchange in asymptomatic congenital heart disease and in prospective, controlled studies of long-COVID. I also developed mathematical tools to model 129Xe MR signals during acquisition and reconstruction. The insights gained from my work underscore the potential for 129Xe gas-exchange MRI biomarkers towards a better understanding of cardiopulmonary disease. My work also provides a way to generate a deeper imaging and physiologic understanding of gas-exchange in vivo in healthy participants and patients with chronic lung and heart disease.

High-resolution fMRI using gradient-echo blood-oxygen-level-dependent (BOLD) contrast is beneficial for the non-invasive study of neural microcircuits. However, the signal spatial specificity of the BOLD contrast severely limits the ability to localize regions of neural activity at the mesoscopic scale in the cortex due to signal contamination from large veins. Phase regression is a venous bias correction technique that uses the correlation between magnitude and phase data in large veins to estimate and supress their contribution to the BOLD signal. This thesis further investigates the performance of phase regression by examining the laminar BOLD signal in human ocular dominance columns. Phase regression removes the venous bias from pial veins and large intracortical veins, while not removing the venous bias from venous vessel sizes within the cortex running parallel to the cortical surface. This thesis demonstrates improved laminar BOLD signal specificity that will be beneficial in future high-resolution laminar fMRI studies.

Water diffusion anisotropy MRI is sensitive to microstructural changes in the brain that are hallmarks of various neurological conditions. However, conventional metrics like fractional anisotropy are confounded by neuron fiber orientation dispersion, and the relatively low resolution of diffusion-weighted MRI gives rise to significant free water partial volume effects in many brain regions that are adjacent to cerebrospinal fluid. Microscopic fractional anisotropy is a recent metric that can report water diffusion anisotropy independent of neuron fiber orientation dispersion but is still susceptible to free water contamination. In this paper, we present a free water elimination (FWE) technique to estimate microscopic fractional anisotropy and other related diffusion indices by implementing a signal representation in which the MRI signal within a voxel is assumed to come from two distinct sources: a tissue compartment and a free water compartment. A two-part algorithm is proposed to rapidly fit a set of diffusion-weighted MRI volumes containing both linear- and spherical-tensor encoding acquisitions to the representation. Simulations and in vivo acquisitions with four healthy volunteers indicated that the FWE method may be a feasible technique for measuring microscopic fractional anisotropy and other indices with greater specificity to neural tissue characteristics than conventional methods.

Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease characterized by chronic airflow obstruction, emphysematous destruction, and airway remodeling. Thoracic CT has previously revealed abnormalities in the small airways, where disease onset is believed to initiate. In previous COPD cohort studies, airway wall thinning and diminished total airway count (TAC) were observed with increasing disease severity. However, longitudinal insights are lacking. Accordingly, the objective of this thesis was to evaluate longitudinal CT airway measurements at baseline and after three-years in ex-smokers. I observed that CT TAC was decreased only in ex-smokers with COPD, whilst airway walls were thinner in both ex-smokers with and without COPD. To my knowledge, this is the first study to show TAC worsening over time in COPD, which suggests airway narrowing, obstruction, and/or obliteration. These longitudinal three-year findings in ex-smokers, in whom forced expiratory volume in 1-second did not change, provide insights into mechanisms of COPD progression.

Patients with chronic kidney disease requiring hemodialysis to sustain life have extremely high rates of cardiovascular morbidity/mortality. This is both a consequence of the disease process and treatment. Hemodialysis induces systemic circulatory stress is largely due to the extracorporeal circuit and volume removal. The stress induced by intermitted hemodialysis is repetitive with cumulative cardiovascular effect and is the principal driver of heart failure and sudden death in this vulnerable population. The insults imposed by hemodialysis result in left ventricular regional wall motion abnormalities and lead to myocardial stunning with permanent damage to the vasculature and myocardium. Currently, the adversity caused by hemodialysis remains under appreciated and largely ignored. The aim of this thesis is to explore intradialytic methods to detect and ameliorate hemodialysis-induced circulatory stress (myocardial stunning). The utility of a non-invasive intradialytic hemodynamic monitoring system using photoplethysmography to detect hemodialysis-induced circulatory stress through the skin was explored. In this first thesis project, we determined that pulse strength, an output generated by PPG technology was associated with the development of myocardial stunning. Building on this initial finding, the second thesis project supported the significance of the pulse strength variable as a signal to the development of regional wall motion abnormalities/myocardial stunning during hemodialysis. Additionally, we determined that intradialytic changes in cutaneous perfusion were detected earlier in treatment and were directly associated with direct measures of global cardiac perfusion using intravenous contract/computerized tomography imaging. Both findings were associated with rates of ultrafiltration/fluid removal during dialysis (previously identified as principal drivers of hemodialysis-induced acute cardiac injury). Exercise pre-conditioning, a phenomenon that is cardio-protective against ischemic-reperfusion injury was explored, in the form of intradialytic exercise. In the third project, we found there to be a reduction in the number of treatment-induced regional wall motion abnormalities at the peak of hemodialysis stress when intradialytic exercise was incorporated into treatment. Moreover, any amount of exercise had an immediate influence with no detrimental effect on treatment tolerability. This work describes the utility and benefit of non-invasive microcirculatory monitoring for the early detection of hemodialysis-induced circulatory stress and demonstrates the pre-conditioning benefit that intradialytic exercise has on the reduction of myocardial stunning in the hemodialysis population

Epilepsy is a degenerative brain disease characterized by abruption of neural activities that result in seizures. The onset of epileptic seizures are usually from a primary source - the epileptogenic foci (EF) which could be distributed to nearby neurons and tissues. Accurate localization of EF is critical in epilepsy cases where drug treatment has failed, and surgery is indicated to resect the EF to alleviate seizure. Typically, hybrid positron emission tomography (PET) and computed tomography (CT) imaging are performed to functionally localize the EF in drug-resistant epilepsy for surgical planning when anatomical abnormalities representing the EF cannot be identified on magnetic resonance imaging (MRI). The recent introduction of integrated PET/MRI scanning has significantly enhanced the localization of EF and eliminated the use of CT for PET attenuation correction (AC), minimizing radiation exposure particularly in radiosensitive patients such as pediatrics. The objective of this thesis is to develop an image analysis approach to further reduce PET radiation dose for optimal pediatric epilepsy PET/MRI. First, to eliminate CT for PET AC, a robust deep learning approach validated in pediatric populations for synthesizing CT from MRI was implemented after performing a rigorous systematic review and meta-analysis of state-of-the art machine learning (ML) AC methods. Next, a deep learning tool using the Self-SiMilARiTy-Aware Generative Adversarial framework (SMART) was developed and evaluated for denoising of PET images acquired with 90% reduction in PET dose, to generate high quality PET images. By combining ML-AC and SMART-PET, this work proposed an approach to drastically reduce the radiation exposure for high quality pediatric epilepsy PET imaging from ~6 mSV in PET/CT to ~0.5 mSV in low-dose PET/MRI.

Lack of an appropriate animal model prevents rapid evaluation of health outcomes associated with hemodialysis. We developed a miniaturized dialyzer for use in small rats. The aim of this study was to investigate whether the miniaturized dialyzers can differentiate performance characteristics representing hemodialysis membrane and fiber design. We evaluated internal filtration, water permeability, sieving coefficient (SC) and clearance of solutes, in the Medium Cut Off (MCO) and conventional High Flux (HF) miniaturized dialyzers. We found statistically significant differences between miniaturized MCO and conventional HF dialyzers with respect to ultrafiltration coefficient, clearance of Beta-2 Microglobulin (B2M), and SC of B2M.These outcomes suggest that the miniaturized dialyzers can effectively differentiate performance characteristics representing hemodialysis membrane and fiber design. This indicates that the miniaturized dialyzers are an appropriate component of a pre- clinical pipeline of dialysis technology impact.

Magnetic resonance imaging (MRI) has cemented itself as the gold standard for imaging of soft tissues and is only increasing in popularity. Given the rising number of MRI scanners and medical device being implanted into patients, it is becoming increasingly likely that patients undergoing MRI will have an implanted medical device (IMD). The presence of an elongated metallic IMD inside a scanner could result in dangerous interactions with the radiofrequency fields during MRI, thus some of these IMDs preclude the patients from being scanned. Orthopedic devices typically fall into this category due to their high potential for RF induced heating, and typically perform poorly in the current standard test method for RF heating. That said, there exists a subset of orthopedic IMDs that still ‘fail’ the current safety testing by heating slightly above the current acceptance criterion. It is hypothesized that such IMDs are not truly a hazard to the patient but are likely failing due to the conservative nature of the current RF heating test (ASTM F2182-19a). In this thesis, novel test platforms are presented for more realistic evaluation of RF heating in orthopedic IMDs, which were used to experimentally challenge the behavior of their simulated counterparts. These test platforms were designed to address the simplifications in the current ASTM test standard that led to exaggerated heating compared to what is expected in patients, namely geometry/material mimicking and perfusion cooling. Heating of a sample implant was simulated (Sim4Life) in these novel test platforms, along with experimental verification of two phantoms to determine agreement with simulation. Simulations (and experimental work) indicated that IMD heating in these realistic phantoms could be anywhere from 20-50% lower than the current ASTM phantom, which is a reasonable estimate of the magnitude of the safety margin involved. It appears perfusion cooling is most effective at reducing IMD heating (compared to geometry/tissue mimicking differences), though improved experimental verification is required before these simulations can influence regulatory change. Introducing empirical evidence of perfusion cooling to regulatory conversations around implant safety would improve access to MRI for the millions living with such marginally unacceptable orthopedics.

Cardiac sodium (Na+) MRI has the potential to unveil relationships between chronic kidney disease (CKD) and cardiovascular disease (CVD). Unfortunately, this modality requires many technical adaptations before it’s possible to extract and compare image data between individuals. This study investigated whether cardiac electrocardiogram (ECG) gating could improve image quality, and whether a uniform phantom could increase signal uniformity if a surface coil is used to acquire the image. Healthy participants were imaged with a custom-built surface coil and 3.0-T MR system. Cardiac gated images presented with a decreased signal-to-noise ratio (SNR), but intensity correction significantly increased image signal uniformity. Future studies using cardiac Na+MRI with a surface coil should consider implementing signal intensity correction but should recognize that cardiac gating reduces image quality unless scan time is increased. Cardiac sodium (Na+) MRI has the potential to unveil relationships between chronic kidney disease (CKD) and cardiovascular disease (CVD). Unfortunately, this modality requires many technical adaptations before it’s possible to extract and compare image data between individuals. This study investigated whether cardiac electrocardiogram (ECG) gating could improve image quality, and whether a uniform phantom could increase signal uniformity if a surface coil is used to acquire the image. Healthy participants were imaged with a custom-built surface coil and 3.0-T MR system. Cardiac gated images presented with a decreased signal-to-noise ratio (SNR), but intensity correction significantly increased image signal uniformity. Future studies using cardiac Na+ MRI with a surface coil should consider implementing signal intensity correction but should recognize that cardiac gating reduces image quality unless scan time is increased.

Introduction. Fractional flow reserve (FFR) is presently an invasive coronary clinical index. Non-invasive CT imaging combined with computational coronary flow modelling may reduce the patient’s burden of undergoing invasive testing. Research statement. The ability to obtain information of the hemodynamic significance of detected lesions would streamline decision making in escalation to invasive angiography. Methods. A reduced order (lumped parameter) model of the coronary vasculature was further developed. The model was used in the assessment of the roles of structure and function on the FFR. Sophisticated methods were used to elicit numerical solutions. Further, CT imaging (n = 10) provided multiple porcine geometries based upon algorithms encoded within an existing scientific platform. Results. It was found that the length of large vessel stenosis and presence of microvascular disease are primary regulators of FFR. Further, the CT data provided a basis to investigate relationships between coronary geometry (structure) and blood flow (function) attributes. Discussion. The presented model, upon personalization, may compliment and streamline ongoing imaging efforts by guiding FFR assessment. It is likely to assist in preliminary data generation for future projects. The computational geometries will contribute to an open source service that will be made available to our University’s researchers.

The aim of this scoping review is to identify, categorize and report current research for the effect and related toxicities of radiotherapy on female erectile tissues.

We propose a multi-institutional effort to define mandibular osteoradionecrosis (MORN) risk modeling via shared development of radiotherapy normal tissue complication probability and prediction model development through dose-inclusive analyses. This analysis will serve as a benchmark model, not only for MORN, but for validated radiotherapy toxicity modeling.

This report is a written protocol for a scoping review of cough sounds in diagnostics. Protocol structure is modeled off of PRISMA-P 2015 guidelines. Cough is a classic symptom in respiratory conditions. Cough is associated with a specific pattern of lower and upper airway vibrations and it carries acoustic information on the condition of the respiratory system. There is an increasing interest in cough sounds as a potential biomarker of health and disease in our setting of increasing data processing capabilities. Studies have demonstrated the existence of acoustically different cough classes in different airway pathologies. However, there are no standard methods for cough sound data collection and analysis. A scoping review is indicated to survey prior approaches and their limitations. This scoping review aims to look at the broad uses of cough sounds in diagnostics, screening, treatment monitoring, cough counting and detection regardless of analytical techniques used, diagnoses investigated, and target populations.

Introduction. Hypoxia – reduced oxygen availability at the tissue level – is a health and incapacitation risk factor when ascending in the atmosphere such as during mountain climbing or in aviation. As a part in the promotion of safety in military aviation, flying personnel are exposed to brief, acute exposures to simulated elevations at regular intervals during their careers, known as hypoxia recognition training. At simulated high-altitude exposures to about 7600 m in hypobaric chambers, the participants may experience severe hypoxia, and the concurrent blood oxygen saturation may drop towards 50 % hemoglobin saturation. During these simulated high-altitude exposures, the health status of the participants is monitored by clinicians assisted by the use of pulse oximeters. Pulse oximeters that are to be approved for medical use are validated at oxygen saturations between 70 and 100 % and using arterial blood gas measurement (SaO2) by hemoximetry. There is thus a relatively large gap between the measurement range used in medical validation experiments and during hypoxia recognition training. Purpose. The overall purpose of this study is therefore to investigate and compare the validity of four different finger pulse oximeters on oxygen saturation and heart rate at very low oxygen saturation levels by reference measurements in arterial blood and electrocardiogram. Methods. Potential participants are initially examined by a medical doctor and descriptive variables are measured. The intervention is carried out in a laboratory with the participants sitting at rest in a dental chair. A total of four separate finger pulse oximeters are examined simultaneously on each of the participants, and the placement of sensors is assigned after randomization. The reference measure for oxygen saturation are taken in arterial blood gas by hemoximetry, and for heart rate electrocardiogram (ECG) is used. The participant will be exposed to normobaric hypoxia using breathing air mixtures containing reduced oxygen concentration supplied by a breathing gas blender. Four to seven different air mixtures are given in a stepwise manner to introduce stable oxygen saturation plateaus of 60–100 %. At stable oxygen saturations, simultaneous samples are taken from the reference measures and the pulse oximeters trice separated by 30 s. The validity is examined across all data and within the saturation plateaus using the Bland & Altmans Limit of Agreement method (LoA). Data from the different manufacturers are also compared. The study will follow the ethical guidelines set by the Declaration of Helsinki, and pre-approval was given by the Regional Committee for Medical and Health Research Ethics, region South-East A. The study will follow General Data Protection Regulation (GDPR) and the local Data Protection Authority was consulted a priori. All volunteers are required to give their written informed consent prospectively.

Developing outcome prediction models for survival, radiation-induced liver disease, radiation-induced liver lymphopenia, and nonlocal failure in advanced hepatocellular carcinoma.

Non-small cell lung cancer (NSCLC) is one of the most diagnosed cancers in Canada, and the leading cause of cancer deaths. A significant challenge in treating NSCLC is balancing aggressive treatment with the potentially severe side effects. In radiation therapy, the management of respiratory motion and the risks of radiation-induced lung injury (RILI) pose significant challenges. 4-dimensional computed tomography (4D-CT) is an important part of motion management, but images often suffer from motion-induced artifacts. Volumetric CT scanners with wide axial field-of-view (aFOV) may reduce these artifacts and present an opportunity to advance CT-based functional lung imaging. Chapter 2 presents a phantom imaging study to investigate the suitability of a 256-slice volumetric CT (vCT) scanner for radiotherapy treatment planning. The density of the highest density materials was under-estimated by the scanner, which can be addressed with the use of an appropriate relative electron density (RED) curve. An average RED curve for all aFOV settings may be used. Chapter 3 presents a study of phantom and NSCLC patient 4D-CT images acquired on a clinical scanner and a vCT scanner. The v4D-CT images were re-sampled to simulate a conventional acquisition using a narrow aFOV clinical scanner. The phantom images demonstrated that target contouring variability decreased in v4D-CT imaging as compared to clinical 4D-CT. In the patient images, mean Hausdorff distance between organs at risk (OAR) contours was significantly correlated to respiratory phase, indicating that motion artifacts contribute to this variability. Chapter 4 presents a novel acquisition and analysis pipeline to image lung ventilation (V), perfusion (Q) and V/Q ratio in a single volumetric CT scan. In a porcine study, these images of V and Q were significantly correlated to standard Xe-enhanced ventilation and PET perfusion images in voxel-wise analysis. In a NSCLC patient study, the images were sensitive to changes in V and Q between baseline imaging and follow-up 6 weeks after radiotherapy. In this thesis, I demonstrate that volumetric CT scanners are suitable for use in radiation therapy simulation and treatment planning, and detail two scanning protocols which may reduce the challenges posed by respiratory motion and RILI risk in NSCLC.

Background: Atrial fibrillation (AF) is a prevalent cardiac disease which has been associated with increased risk of dementia and cognitive decline. We hypothesize that atrial fibrillation leads to regional transient hypoperfusion events in the brain, and that geometric variations in the arterial structure called the Circle of Willis (CoW) play a role in these events. Methods: A computational model was developed to simulate cerebral blood flow in six common variations of the CoW. Risk was assessed based on frequency of beat-wise regional hypoperfusion events during AF, and sensitivity analysis was performed with respect to this model output. Results: A key artery in the CoW, called the A1 segment, was found to play the most important role in cerebral perfusion. Intrinsic heart rate was also found to influence the frequency of hypoperfusion events. Conclusions: Our results suggest that heart rate and CoW geometry play important roles influencing cerebral hemodynamics during AF.

Near Infrared Spectroscopy (NIRS) is a clinical non-invasive instrument that provides insight regarding the relative or absolute tissue oxygenation with respect to adequate delivery of oxygen and metabolic demands. The technique capitalizes on the light absorbance properties of heme-containing groups, hemoglobin and myoglobin, with the aim to assess the microvasculature oxygenation kinetics in many clinical populations. NIRS relies on two principles: a) tissue is relatively transparent to near-infrared light and b), the tissue has compounds such as hemoglobin in which light absorption depends on the oxygenation status of the tissue. By measuring light propagation in tissue, NIRS is able to detect changes in the relative concentration of different light-absorbing molecules. There are three main categories of near-infrared light spectrometers: continuous wave, time domain and frequency domain spectrometers. This scoping review is one component of a larger series of reviews assessing perfusion, blood flow, and vascular structure in order to provide guidance for translational research scientists. This review examines 50 studies retrieved from a larger search of databases Embase and Medline. This review provides an overview of the pathophysiology underlying NIRS, the equipment and protocol for measuring NIRS, and provide a scoping review for the clinical and translational research applications of NIRS.

Blood-mimicking fluids (BMFs) are often used to investigate blood flow using physical replicas of vessels with cardiovascular disease. Particle image velocimetry (PIV) is used with silicone poly-dimethyl siloxane (PDMS) vascular models to visualize this flow. The challenge is creating a blood-mimicking fluid that matches the viscosity profile, viscoelasticity, and density of whole blood while also matching the refractive index (RI) of PDMS. Water-glycerol solutions are commonly used with sodium iodide (NaI) added to increase the RI without changing viscosity. However, NaI is expensive, stains easily, and turns fluids from optically clear to yellow in less than a day. Furthermore, without adding a shear-thinning agent such as xanthan gum, BMFs do not mimic the aggregation of red blood cells at low arterial shear rates. This study characterized 12 published BMF formulations with varying concentrations of water, glycerol, xanthan gum and NaI/urea. Rheological measurements were obtained using the Anton Paar MCR 302, RI was measured using an Abbe Refractometer, and density was found using a 10 ml pipette and analytical balance. It was found that solutions with 0.04 wt.% xanthan gum and a low ratio of glycerol to water best match the shear-thinning profile and viscoelasticity of whole blood over a wide range of shear rates. As well, urea is a cheaper alternative to NaI, as it can match the density, RI of PDMS, and viscosity of blood while maintaining a transparent colour. Overall, it was concluded that a blood-mimicking fluid with 51.9 wt.% water, 22.2 wt.% glycerol, 25.9 wt.% urea/NaI, and 0.04 wt.% xanthan gum was the best candidate for investigating blood flow with PIV.

Microvascular mechanisms that regulate the microcirculation can be observed within blood flow as coupled oscillations operating over a wide range of different frequencies and time scales. Abnormally discordant interactions between regulation mechanisms can cause derangement of microvascular flow patterns, which has been suggested to be indicative of increased cardiovascular disease risk. Based on the premise that interactions between microvascular mechanisms are key to understanding flow regulation and disease progression, wavelet-based phase coherence was used to characterize the relationship between cardiac, respiratory, myogenic, neurogenic, and metabolic regulation across different microvascular beds and physiological conditions. Steady state cutaneous microvascular blood flow data was collected using laser Doppler flowmetry in two locations: the forearm and forehead. At both locations, blood flux data was recorded under two hemodynamic steady states: at rest (30° C) and after local skin warming (45° C). The wavelet transform was used to calculate wavelet phase coherence between forehead and forearm blood flux, which was compared for low vs. high temperature hemodynamic states using frequency ranges associated with specific regulation mechanisms. Metabolic, neurogenic, and endothelial NO-dependent regions demonstrated significant difference in time-averaged wavelet phase coherence, with greater coherence in the high temperature state. Time-localized wavelet phase coherence was relatively consistent across the 15-minute data sample. Study results favorably demonstrate the ability of wavelet-based phase coherence to differentiate between physiological conditions and observe disease progression; characterizing the relationship between different regulation mechanisms with coupling functions via Bayesian Inference are of interest in future work.

Intratumoral modulation therapy (IMT) is a novel electrotherapy used to treat brain cancer tumours using electric fields applied directly to the tumours through implanted electrodes. Previous research has validated IMT's effectiveness and provided computer-simulated optimizations for IMT electric fields. This work validates these computer optimizations in-vitro, using a PCB construct to deliver electric fields, and bioluminescence imaging to assess cell viability. We found electric field strength to correlate with cell viability, and found that rotating (phase-shifted) electric fields did not produce significant improvements in IMT efficacy. Future work will investigate different IMT frequencies and other parameters, while providing biological replicates to strengthen our results.

Successful segmentation of the total intracranial vault (ICV) and ventricles is of critical importance when studying neurodegeneration through neuroimaging. We present iCVMapper and VentMapper, robust algorithms that use a convolutional neural network (CNN) to segment the ICV and ventricles from both single and multi-contrast MRI data. Our models were trained on a large dataset from two multi-site studies (N=528 subjects for ICV, N=501 for ventricular segmentation) consisting of older adults with varying degrees of cerebrovascular lesions and atrophy, which pose significant challenges for most segmentation approaches. The models were tested on 238 participants, including subjects with vascular cognitive impairment and high white matter hyperintensity burden. Two of the three test sets came from studies not used in the training dataset. We assessed our algorithms relative to four state-of-the-art ICV extraction methods (MONSTR, BET, Deep Extraction, FreeSurfer), as well as a ventricular segmentation tool (FreeSurfer). Our multi-contrast models outperformed other methods across all evaluation metrics, with average Dice coefficients of 0.98 and 0.94 for ICV and ventricular segmentation respectively. Both models were also the most time efficient, segmenting the structures in orders of magnitude faster than some of the other available methods. Our networks showed an increased accuracy with the use of a conditional random field (CRF) as a post-processing step. We further validated both segmentation models, highlighting their robustness to images with lower resolution and signal-to-noise ratio, compared to tested techniques. The pipeline and models are available at: https://icvmapp3r.readthedocs.io and https://ventmapp3r.readthedocs.io to enable further investigation of the roles of ICV and ventricles in relation to normal aging and neurodegeneration in large multi-site studies.

Minimally invasive procedures for prostate cancer diagnosis and treatment, including biopsy and brachytherapy, rely on medical imaging such as two-dimensional (2D) and three-dimensional (3D) transrectal ultrasound (TRUS) and magnetic resonance imaging (MRI) for critical tasks such as target definition and diagnosis, treatment guidance, and treatment planning. Use of these imaging modalities introduces challenges including time-consuming manual prostate segmentation, poor needle tip visualization, and variable MR-US cognitive fusion. The objective of this thesis was to develop, validate, and implement software- and hardware-based tools specifically designed for minimally invasive prostate cancer procedures to overcome these challenges. First, a deep learning-based automatic 3D TRUS prostate segmentation algorithm was developed and evaluated using a diverse dataset of clinical images acquired during prostate biopsy and brachytherapy procedures. The algorithm significantly outperformed state-of-the-art fully 3D CNNs trained using the same dataset while a segmentation time of 0.62 s demonstrated a significant reduction compared to manual segmentation. Next, the impact of dataset size, image quality, and image type on segmentation performance using this algorithm was examined. Using smaller training datasets, segmentation accuracy was shown to plateau with as little as 1000 training images, supporting the use of deep learning approaches even when data is scarce. The development of an image quality grading scale specific to 3D TRUS images will allow for easier comparison between algorithms trained using different datasets. Third, a power Doppler (PD) US-based needle tip localization method was developed and validated in both phantom and clinical cases, demonstrating reduced tip error and variation for obstructed needles compared to conventional US. Finally, a surface-based MRI-3D TRUS deformable image registration algorithm was developed and implemented clinically, demonstrating improved registration accuracy compared to manual rigid registration and reduced variation compared to the current clinical standard of physician cognitive fusion. These generalizable and easy-to-implement tools have the potential to improve workflow efficiency and accuracy for minimally invasive prostate procedures.

Joint replacements are becoming increasingly commonplace with over 130,000 joint arthroplasties being performed annually in Canada. Although joint replacement surgery is highly successful, implants do occasionally fail and need to be replaced via costly and difficult revision surgery. Periprosthetic joint infection (PJI) has recently become the leading reason for revision of both hip and knee replacements, which is unfortunate because PJI is difficult to diagnose and treat effectively; diagnosis is made particularly difficult by the lack of established non-invasive (imaging) means of evaluating PJI. This thesis aims to demonstrate that magnetic resonance imaging (MRI) has potential for diagnosing and monitoring PJI through advances in implant design and novel application of quantitative imaging. The recent proliferation of metal 3D-printing has already inspired the clinical use of 3D-printed porous metal devices due to their favorable osseointegration and mechanical properties. This thesis explores an important MRI benefit to porous implants: their decreased effective magnetic susceptibility and proportional decrease in imaging artifacts. This is relevant to PJI because MRI is already well-established in diagnosing musculoskeletal infections, but metals cause image obscuring signal loss. This work shows that 3D-printed porous metal structures are likely to avoid this limitation, as their effective magnetic susceptibility is linearly proportional to porosity; if true, MRI will be able to diagnose PJI as easily as non-prosthetic joint infections. This thesis describes a novel use for two important parameters measured by quantitative MRI: effective relaxation rate (R2*) and magnetic susceptibility (QSM; quantitative susceptibility mapping). This work seeks to address an important unmet need in PJI treatment – the ability to monitor drug release during localized antibiotic delivery – by exploiting these parameters’ proportionality to gadolinium concentration. This idea is centered around using gadolinium-based MRI contrast agents as a surrogate small-molecule that acts as a proxy for drugs to study diffusion-controlled release. An initial implementation of this concept showed promising results, including the ability to fit the data to a mathematical model of drug release. This shows the potential of MRI as a non-invasive means of monitoring localized antibiotic treatment of PJI post-revision.

The long-term neuropsychological, cognitive, and neurobiological effects of the severe acute respiratory syndrome (SARS-CoV-2) in survivors with milder symptoms are still poorly understood. In this thesis we evaluated cognitive and psychological changes approximately five weeks after a wide range of symptoms in COVID-19 illness and determined whether advanced diffusion magnetic resonance imaging measures within subcortical brain structures of the limbic system were related to neurological, respiratory, psychiatric, and gastric symptoms experienced during the acute phase of illness. Cognitive and neuropsychological evaluations were performed in 45 participants who experienced neurological symptoms during the acute phase of COVID-19 illness. Participants also underwent a 7 Tesla MRI neurological exam on the same day. The group showed a significant reduction in attention compared to a normative population, but no differences in other cognitive domains. Although white matter hyperintensities were visible on fluid attenuated inversion recovery (FLAIR) images in 22 of 43 participants consistent with small vessel ischemic disease and migraine, this incidence is consistent with that expected in a normative population. Participants were divided into groups based on the presence or absence of symptoms at their acute illness from the medical history collected over the phone or in-person during recruitment. No differences were observed in subcortical brain structure volumes when comparing participants between subgroups. Differences in advanced diffusion metrics were observed within several subcortical structures (p

Magnetic resonance imaging utilizes electromagnets to produce anatomical images in both clinical and research settings. In the race towards increasing performance head-optimized scanners have begun playing a significant role in providing high quality imaging of the head. However, they are implemented using smaller geometries and as such fail to allow entrance of the patient past their shoulders. This is overcome by designing asymmetric gradient coils which have their imaging region located towards one end of the gradient coil, as opposed to the geometric center, allowing brain imaging. There exists interest in compact configurations which allow imaging further into the cervical spine which is unfeasible using current asymmetric gradients. This work seeks to explore the design of asymmetric gradient coils with shoulder cut-outs to enable neck imaging by allowing the patient to enter further into the gradient coil while maintaining the small inner radius of a head-only platform. First, the relative trade-offs in designing an asymmetric shoulder cut-out gradient coil are explored and extended by rotating the transverse gradient axes to produce gradient coils which compensate for some of the electromagnetic burden due to the loss of conducting surfaces on the sides. Next, a complementary set of spherical harmonic active shims are designed and explored for implementation within this gradient coil configuration. From there the design of a cylindrical radiofrequency coil using gradient design techniques is investigated as preliminary work towards implementing these low-frequency design techniques which have had success designing gradient coils towards the design of radiofrequency coils. Finally, motivated by the complexity of the induced eddy currents in the surrounding conductive structures due to asymmetric gradient coils the final project explores the design of a multi-coil matrix array aimed at fitting within the compact gradient housing to dynamically compensate eddy currents during imaging. This work ultimately demonstrates the feasibility of implementing an asymmetric shoulder cut-out gradient coil with rotated transverse gradient axes to enable neck imaging in a compact MRI scanner while providing potential solutions to handle the increased eddy current complexity associated with a setup such as this.

Background/Objective: Structural brain magnetic resonance imaging (MRI) is not mandatory in Alzheimer's disease (AD) research or clinical guidelines. We aimed to explore the use of structural brain MRI in AD/mild cognitive impairment (MCI) trials over the past 10 years and determine the frequency with which inclusion of standardized structural MRI acquisitions detects comorbid vascular and non-vascular pathologies. Methods: We systematically searched ClinicalTrials.gov for AD clinical trials to determine their neuroimaging criteria and then used data from an AD/MCI cohort who underwent standardized MRI protocols, to determine type and incidence of clinically relevant comorbid pathologies. Results: Of 210 AD clinical trials, 105 (50%) included structural brain imaging in their eligibility criteria. Only 58 (27.6%) required MRI. 16,479 of 53,755 (30.7%) AD participants were in trials requiring MRI. In the observational AD/MCI cohort, 141 patients met clinical criteria; 22 (15.6%) had relevant MRI findings, of which 15 (10.6%) were exclusionary for the study. Discussion: In AD clinical trials over the last 10 years, over two-thirds of participants could have been enrolled without brain MRI and half without even a brain CT. In a study sample, relevant comorbid pathology was found in 15% of participants, despite careful screening. Standardized structural MRI should be incorporated into NIA-AA diagnostic guidelines (when available) and research frameworks routinely to reduce diagnostic heterogeneity.

Biocompatible titanium-alloys can be used to fabricate patient-specific medical components using additive manufacturing (AM). These novel components have the potential to improve clinical outcomes in various medical scenarios. However, AM introduces stability and repeatability concerns, which are potential roadblocks for its widespread use in the medical sector. Micro-CT imaging for non-destructive testing (NDT) is an effective solution for post-manufacturing quality control of these components. Unfortunately, current micro-CT NDT scanners require expensive infrastructure and hardware, which translates into prohibitively expensive routine NDT. Furthermore, the limited dynamic-range of these scanners can cause severe image artifacts that may compromise the diagnostic value of the non-destructive test. Finally, the cone-beam geometry of these scanners makes them susceptible to the adverse effects of scattered radiation, which is another source of artifacts in micro-CT imaging. In this work, we describe the design, fabrication, and implementation of a dedicated, cost-effective micro-CT scanner for NDT of AM-fabricated biomedical components. Our scanner reduces the limitations of costly image-based NDT by optimizing the scanner's geometry and the image acquisition hardware (i.e., X-ray source and detector). Additionally, we describe two novel techniques to reduce image artifacts caused by photon-starvation and scatter radiation in cone-beam micro-CT imaging. Our cost-effective scanner was designed to match the image requirements of medium-size titanium-alloy medical components. We optimized the image acquisition hardware by using an 80 kVp low-cost portable X-ray unit and developing a low-cost lens-coupled X-ray detector. Image artifacts caused by photon-starvation were reduced by implementing dual-exposure high-dynamic-range radiography. For scatter mitigation, we describe the design, manufacturing, and testing of a large-area, highly-focused, two-dimensional, anti-scatter grid. Our results demonstrate that cost-effective NDT using low-cost equipment is feasible for medium-sized, titanium-alloy, AM-fabricated medical components. Our proposed high-dynamic-range strategy improved by 37% the penetration capabilities of an 80 kVp micro-CT imaging system for a total x-ray path length of 19.8 mm. Finally, our novel anti-scatter grid provided a 65% improvement in CT number accuracy and a 48% improvement in low-contrast visualization. Our proposed cost-effective scanner and artifact reduction strategies have the potential to improve patient care by accelerating the widespread use of patient-specific, bio-compatible, AM-manufactured, medical components.

While prostate cancer is not typically an aggressive disease, it can be lethal for some patients. Treatment of the disease has proven to be effective; however, when the disease recurs locally it does so most commonly at the site of the dominant intraprostatic lesion (DIL). An effective treatment method for prostate cancer is high dose rate brachytherapy (HDR-BT). This is a radiation therapy where a radioactive seed is interstitially inserted into the patient to deliver the radiation dose. It was the goal of this work to utilize preprocedural imaging and pathological data from radical prostatectomy specimens to investigate methods to dose escalate DILs. The first objective was to establish a baseline range of doses that DILs would receive within clinically standard whole-gland HDR-BT treatment plans. To execute this objective, DILs were retrospectively registered within previously implemented HDR-BT plans to analyze DIL dose. The next objective was to investigate targeting mpMRI and PSMA-PET DILs and to determine if this would translate to escalated dose to the pathologically defined cancer. This was performed by retrospectively developing mpMRI and PSMA-PET DIL targeted treatment plans and comparing the doses delivered to co-registered digital histology between targeted and whole-gland plans. The final investigation that was performed aimed to determine an optimal margin to be added to the mpMRI DILs to increase dose to the pathologically confirmed cancer . To do so, we created mpMRI targeted HDR-BT treatment plans with varying margins added to the DIL and analyzed the dose that would have been delivered to the pathologically confirmed cancer. The results of this work demonstrated that while most DILs would receive the prescription dose during HDR-BT, up to 26% may receive less than the intended dose . Secondly, it was shown that by targeting predefined DILs on mpMRI or PSMA-PET with a boost of 130% of the 15-Gy prescription resulted in an average increase of 1.3 Gy to the pathologically defined high-grade disease in comparison to standard whole-gland plans. Lastly, there is the potential to further increase the dose to the pathologically confirmed disease by adding an expansion margin of 1 mm to the mpMRI-defined DILs.