oURspace (Univ. of Regina)
Not a member yet
15455 research outputs found
Sort by
Advanced CNN architecture integrating machine learning algorithms for precise Alzheimer's disease classification
No full textA Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Applied Science in Industrial Systems Engineering, University of Regina. xiii, 158 p.Alzheimer's disease (AD) is a degenerative neurological disorder that affects millions of individuals worldwide and is very difficult to detect and treat in its early stages. This thesis presents a novel architecture for a convolutional neural network (CNN) designed exclusively to classify Alzheimer's disease using functional magnetic resonance imaging (fMRI) data. This work improves the accuracy and reliability of early Alzheimer's identification by using state-of-the-art deep learning techniques to the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset.
The basis of this research is the ADNI dataset, a vast collection of brain imaging and associated data from people with different degrees of cognitive impairment. The primary objectives are to classify Alzheimer's disease into distinct categories using cognitively normal (CN), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI), and Alzheimer's disease (AD) using the recently developed CNN architecture. This study also uses transfer learning techniques to compare the performance of the new CNN with well-known deep learning models like ResNet50 and VGG16, as well as with more conventional machine learning algorithms like XG Boost, k-nearest neighbor (KNN), and Random Forest.
The innovative CNN architecture is meticulously designed to maximize classification accuracy. The preprocessing steps involve resizing fMRI images to 109x91 pixels and labeling them accordingly. The network incorporates convolution layers with 3x3 kernels, ReLU activation functions, and 2x2 pooling layers, transforming the images into feature vectors that are subsequently classified. Compared to previous tested models, the innovative CNN architecture performed better, achieving an impressive 99.51% classification accuracy. In terms of comparison analysis, the accuracy of the VGG16 model was 98.24%, whereas the accuracy of the ResNet50 model was 96.05%. The XG Boost classifier, combined with VGG16 for feature extraction, reached an accuracy of 96.93%. The KNN algorithm, also paired with VGG16, exhibited outstanding performance with an accuracy of 98.68%, making it the most effective among the traditional machine learning methods tested. With VGG16 included, the Random Forest classifier produced an accuracy of 94.70%.
The outcomes demonstrate how well the suggested CNN architecture performs in comparison to current deep learning and machine learning models in precisely classifying Alzheimer's disease stages. This study demonstrates how sophisticated CNN designs and transfer learning can be used to enhance Alzheimer's disease early detection and diagnosis. The findings suggest that further exploration of alternative deep learning networks, such as convolutional auto encoders, Alex Net, and Google Net, as well as ensemble methods, could enhance model generalization and minimize overfitting.
In conclusion, this thesis presents a significant advancement in Alzheimer’s disease classification using fMRI data, providing a robust framework for future research and development in neuroimaging and deep learning applications. The superior performance of the novel CNN architecture demonstrates its potential as a valuable tool for early diagnosis, which is crucial for managing and potentially mitigating the way Alzheimer's disease advances.Studentye
Geochemical modeling of diagenesis, hydrothermal alteration, and unconformity-related uranium mineralization in the Athabasca Basin, Saskatchewan, Canada
No full textA Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy in Geology, University of Regina. xvi, 203 p.Unconformity-related uranium (URU) deposits in the Athabasca Basin were interpreted to have formed through interactions between oxidizing basinal fluids and reducing basement-derived fluids or basement lithologies under diagenetic-hydrothermal conditions. However, there is controversy regarding whether U is ultimately derived from the basin or the basement, fluid flow mechanisms responsible for metal leaching and transport, and deposition mechanisms of ores. This study uses geochemical modeling to: 1) determine fluid flow patterns responsible for U leaching and transport in the Athabasca basin; 2) recognize critical factors controlling ore deposition near the unconformity intersected by basement faults; and 3) constrain metal sources, fluid migration pathways, and ore precipitation mechanisms for U and associated elements, including Ni, Co, and As, which are anomalously enriched within some URU deposits.
Petrographic observation on four drill cores indicates that the top of the sandstone succession below a mud-rich aquitard is characterized by extensive quartz overgrowths, whereas the basal part contains little cement and shows extensive dissolution features. Reactive transport modeling indicates that such a quartz cementation-dissolution pattern can be produced only if the sandstones are sufficiently permeable and thick so that the Rayleigh number exceeds the critical value for thermal convection. The results indicate that thermal convection did occur in the Athabasca Basin and may have facilitated the large-scale circulation of diagenetic fluids to leach and transport U within sandstones.
Reactive transport modeling further shows that significant URU mineralization occurs at a fault-unconformity intersection only if thermal convection and basin-
basement fluid mixing take place concurrently. If there is no thermal convection in the basin, only sparse U mineralization occurs along the unconformity. If insufficient amount of reducing fluid is provided from the basement fault, no significant U mineralization occurs either. Furthermore, no significant U mineralization occurs if the U concentration in the basinal fluid is low. It is concluded that the formation of URU deposits is the result of coupling of three critical factors: high-permeability sandstone favoring thermal convection in the basin, ample supply of reducing fluids along reactivated basement faults, and abundant U-rich basinal fluids in the basin sequences.
Thermodynamic modeling indicates that significant amounts of U can only be transported by highly oxidizing fluids, whereas Ni, Co, and As can be co-transported with U in the same highly oxidizing fluids, or in moderately oxidizing fluids without U. Reaction path modeling further shows that uraninite precipitates before Ni-Co arsenides and sulfarsenides, when ore fluids interact with basement lithologies or mix with reducing fluids. These results confirm the ore precipitation sequences observed in typical URU deposits, and the significance of fluid mixing in ore deposition, and provide theoretical support for crystalline basement rocks as the primary Ni-Co-As source.
The thesis concludes that the basin is the primary U source, and thermal convection is vital for leaching of U from the basin and its transport to mineralization sites; the spatial-temporal coupling of thermal convection and basin-basement fluid mixing is key for U deposition and accumulation at fault-unconformity intersections; Ni, Co, and As were leached from basement rocks by percolating basinal brines, and different patterns of fluid flow and fluid mixing result in the co-mineralization of U, Ni, Co, and As in certain deposits, but not in others.Studentye
The role of municipal decision-making in community wellbeing in times of drought: a case study of Canada’s Sunshine Coast Regional District
No full textA Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Public Policy, University of Regina. xii, 67 p.Access to clean water is essential for meeting personal, societal, and environmental needs. Climate change is causing more frequent and extreme weather events, leading to less dependable water access and intensified droughts (Intergovernmental Panel on Climate Change, 2023; UNICEF, 2023). This poses significant threats to human health, environmental sustainability, and economic prosperity (Brandes & O’Riordan, 2014; United Nations, 2024).
This research explores the role of municipal decision-making during the 2022 drought in the Sunshine Coast Regional District (SCRD), focusing on the impact of water policies on community wellbeing. The study reviews secondary sources, including provincial documents and public reports, and analyzes online questionnaires completed by SCRD’s municipal representatives. Qualitative and quantitative data were analyzed using a convergent parallel design.
Findings indicate low levels of awareness about water problems, usage practices and general water literacy in the region, highlighting the need for behavioural changes in water consumption, conservation, and management. During the 2022 drought, municipalities primarily employed information-based policies and Water Conservation Regulations. Efforts also focused on expanding and diversifying water supply sources. Significant progress in adaptive governance could be achieved and supported by adjusting policies and practices proactively. Participants emphasized the need to consider human, natural, economic, and social aspects of community wellbeing, advocating for a holistic approach to enhance municipal decision-making.
The value a community places on water is intricately connected to how the resource is managed. Municipal decision-making plays a crucial role in delimiting the perspectives and ideas that can lead to a more sustainable and collaborative community. This research contributes to the discourse on sustainable communities and the effectiveness of governmental efforts in addressing droughts.
Keywords: Droughts, community wellbeing, decision-making, water policies, adaptive governanceStudentye
Exploring the mental health, suicidal behaviours, and quality of life among women with fertility problems.
No full textIntegrating stewardship and resource recovery: A dual-faceted analysis of e-waste and used oil management practices of Canadian provinces
No full textA Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Applied Science in Environmental Systems Engineering, University of Regina. ix, 82 p.Canada faces significant challenges in waste management, driven by high per capita waste generation. To address these issues, the country has implemented various waste management stewardship programs aimed at improving waste collection and resource recovery. This study examines the crucial role of stewardship in managing e-waste and used automotive resources including used oil, filters and containers. By focusing on stewardship practices, it highlights how effective management can improve collection rates, enhance resource recovery, and strengthen financial performance. The analysis emphasizes the importance of stewardship approaches to handling these special waste types, illustrating their potential to reduce environmental impact while optimizing resource use across Canadian provinces. The first part presents a comprehensive analysis of e-waste collection and management trends across six Canadian provinces, focusing on e-waste collection rates, provincial stewardship model attributes, program strategies and budget allocations from 2013 to 2020. Temporal and regression analyses were conducted using data from Electronic Product Recycling Association reports. The analysis emphasizes the significant impact of stewardship model attributes on e-waste collection rates, with Quebec emerging as a standout case, showcasing a remarkable 61.5% surge in collection rates. Findings from group analysis reveal a positive correlation between per capita e-waste collection rate and the growth of businesses and collection sites in Western Canada. This highlights the potential benefits of a coordinated waste management approach, emphasizing the importance of shared resources and collaborative policies. Financial aspects of e-waste management are also explored, revealing opportunities for improvement in Saskatchewan and Manitoba, where average allocations to e-waste collection efficiency stand at 6.6% and 7%, respectively. A 40.5% decrease in e-waste collection rates was observed in British Columbia, indicating additional public awareness campaigns may be required, as an 8% decline in consumer outreach was observed during the study period. The first part recommends leveraging region-specific needs to establish a collaborative approach, enhancing e-waste collection efforts. The second part addresses a gap in evaluating the recovery management systems for used oil, filters, and containers. The performance of resources recovery was examined in four Canadian provinces from 2010 to 2022 within automobile industry. The collection rates of resources, financial performance, and temporal changes of two original indicators: Resource Recovery Per Vehicle (RRPV), and Expenses Per Vehicle (EXPV) were examined. British Columbia and Quebec had the highest collection rates of used oil, filters, and containers (mean ranging 83.0 to 92.9%). Despite having the lowest mean collection rate of used oil (71.0%) and filters (78.7%), Saskatchewan has significant RRPV for used oil (20.4 liters) and filters (2.12 units). Decreasing RRPV (-0.01 to -0.38) trends were identified in all jurisdictions, suggesting the need for targeted recovery strategies towards automotive sectors. A mild increasing trend of EXPV in all jurisdictions is observed (slope +0.02 to +0.08). Quebec exhibited the most efficient resource recovery, with EXPV ranging from CAD 2.4 to CAD 3.3 per unit vehicle. Profit margin analysis revealed consistently high margins of 8.6% in Quebec, contrasting with Manitoba's lower 1.32%. The lower profit margin may partly be due to higher administrative costs (16.2%). The findings highlight the potential benefits of the proposed RRPV and EXPV indicators in evaluating management systems for used oil, filters, and containers.Studentye
Wellbeing and protective factors in parents of typically developing young children
No full textA Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Bachelor of Arts (Honours) in Psychology, University of Regina. 52 p.Background: While parents of young children experience challenges to their wellbeing, there is limited research investigating potential protective factors. This study explored the association between potential protective factors (i.e., distress tolerance, emotion regulation, self-efficacy, resilience, and perceived social support) and wellbeing in parents of young, typically developing children.
Methods: Participants included 99 parents (92.9% female, MParent Age = 32.95, SD = 5.134) of young (MChild Age = 24.46months, SD = 15.38), typically developing children recruited in Canada. Participants completed an online questionnaire consisting of demographics, wellbeing, distress tolerance, emotion regulation, self-efficacy, resilience, and perceived social support.
Results: Significant associations were observed between wellbeing and all protective factors (p < .01). Results from linear multiple regression demonstrated that the model accounted for 41.6% of the variance in wellbeing F = (6, 98) = 12.65, p < .001, with emotion regulation (p < .05) and social support (p < .05) being significant predictors.
Conclusions: Relationships exist between wellbeing and protective factors in parents of young, typically developing children. Protective factors account for a large proportion of the variance in parent wellbeing. Impact: The findings highlight potential contributory factors to parent wellbeing. As such, findings identify factors that may represent important targets (i.e., emotion regulation and social support) for programs or interventions focused on supporting and/or bolstering parent wellbeing
Canadians’ opioid awareness: an analysis across multiple demographics
No full textA Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Bachelor of Arts (Honours) in Psychology, University of Regina. 39 p.Canada is currently experiencing an opioid crisis that leads to many Canadian deaths each year. The present study is a quantitative analysis of data collected through Statistics Canada’s Survey on Opioid Awareness (2017). Participants (N= 5,116) answered questions related to their awareness of several topics including: the opioid issue in Canada, their personal opioid use, appropriate overdose response, sharing opioids, harm reduction services, and general information related to opioid use. These topics have been divided into 2 main scales: Awareness of Safe Opioid Use and Awareness of General Information Related to Opioid Use (6 items; r = 0.87). Factor analysis revealed 3 subscales within the first scale, Awareness of Safe Opioid Use. These subscales are Awareness of Appropriate Overdose Response (4 items; r = 0.75), Awareness of Related to Sharing Opioids (3 items; r = 0.63), and Awareness of Harm Reduction Services (3 items; r = 0.69). 3 in 10 participants reported using opioids in the past five years. The majority of those who used opioids were female (57.10% female vs. 42.90% male). Women also report being more aware of appropriate overdose response (54.41% female vs. 45.59% male). 80% of participants reported being at least somewhat aware that there is currently an opioid issue in Canada. Participants over the age of 80 are significantly less aware of general information related to opioid use and appropriate overdose response than most age groups. Residents of Quebec report significantly less opioid use in last five years as well as lower levels of awareness of general information related to opioid use than most provinces. Residents of British Columbia reported being significantly more aware of general information related to opioid use and appropriate overdose response. These findings provide insight into which Canadian populations have the greatest need for information related to opioid use and overdose
If Rocks Could Talk: Laser ablation mapping of 2.5 billion year old banded iron formations
No full textThe development of an efficient model using deep learning for stray clays position prediction in the GPR data for potash mining operations
No full textA Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Applied Science in Electronic Systems Engineering, University of Regina. xv, 91 p.This thesis focuses on the analysis of ground penetrating data obtained from potash mining room roofs with the goal of enhancing safety through the automatic detection of anomalies in the roofs. Ground Penetrating Radar (GPR) is a tool used to detect the position of geological layers of the earth. This is possible by transmitting an electromagnetic (EM) signal through an antenna and capturing the reflected signal representing the positions of the layers. Conventionally, a region of at least 60 centimeters from the roof to the 414 clay seam is necessary for mining operations. Therefore, detecting the position of the 414 clay seam is important. However, stray clays are barriers since they can attenuate the reflected signal from the 414 clay seam, hence limiting our ability to detect the returned signal. The focus of this research is to present a model for detecting stray clays and the 414 clay seam. GPR is an important part of this research for providing data for developing deep learning models to predict the position of the layers. In this study, I used traditional Artificial Neural Networks (ANNs) and Convolutional Neural Networks (CNNs) as two distinct types of models to predict the anomalies in the potash mine. Considering the GPR signal A-scan as the input image for developing the proposed CNN is one of the main contributions in this study. The presented CNN model provides a method to incorporate spatial information facilitating the peak detection of the barriers known as stray clays. The deep learning model has been trained and evaluated on actual GPR data based on the geophysicist’s interpretation as the ground truth data. For easier implementation of the deep learning models, stray clays are divided into stray clay 1 and stray clay 2. Two different mines’ GPR data are used in this study: Cory and Vanscoy. For Cory mine data, the designed ANN provides a training accuracy of 98%, 92%, and 96% for 414 clay seam, stray clay 1, and stray clay 2, respectively. The figures for the test accuracy are 93%, 83%, and 91%. The proposed CNN achieved 98%, 99%, and 98% training accuracy, and
92%, 84%, and 89% test accuracy. Similar results are obtained for the Vanscoy mine dataset. There are various test cases designed for Vanscoy data, where the best training accuracy of 98% and 96%, and test accuracy of 95% and 72% for 414 clay seam and stray clay 1 are obtained, respectively. It is to be noted that unlike the Cory mine, in the Vanscoy mine, there is only stray clay 1 present. The performance of the presented ANN and CNN models is demonstrated by accurately distinguishing between stray clays and the 414 clay seam—a task that has traditionally required a lot of time and resources from geophysicists. The obtained results confirm the effectiveness of the presented prediction models used in this study.
Keywords: Ground Penetrating Radar (GPR), Artificial Neural Network (ANN), Convolutional Neural Network (CNN), Clustered Ratio Derivative (CRD), Potash mine safety, Stray Clays detection, Auto-picking algorithms, Machine Learning, and Deep Learning.Studentye
Prairie mountain biking: A mix methods approach to understanding family participation
No full textinfo-graphicMountain biking is a lifestyle and adventure sport (Immonen, et al., 2017) that “is an exciting, intense, physically challenging sport” (Dodson, 1996, p. 317). Athletes ride bicycles off-road on trails through fields and forests, mountains, and deserts (Moularde & Weaver, 2016). Multiple factors influence youth participation (Bogage, 2017) in organized sport (Jellinek & Durant, 2004). Parents or guardians facilitating youth sport participation (Harwood & Knight, 2015). Mountain bike participation tends to rely upon grassroots organizations to advocate for access to facilities and resources (Buning & Lamont, 2021). Therefore, this study sought to inform Community Sport Organizations (CSO) about their participants’ motivations, desired club elements/offerings and interest in being engaged/involved (e.g., volunteering). The focus is on mountain biking from the CSO members’ perspectives.Offroad Syndicate Mountain Bike Clu