48080 research outputs found
Sort by
Evaluating the Robustness of Intracranial Single-Isocentre Multiple Target Stereotactic Radiosurgery under Varying Treatment Planning Optimisation Methods
No full textSingle Isocentre Multiple Targets (SIMT) Stereotactic Radiosurgery (SRS) is a relatively new radiotherapy technique for treating multiple lesions in a single treatment session. However, the accuracy and robustness of SIMT SRS treatment plans can be affected by various treatment planning and delivery errors. Implementing effective and robust treatment planning optimisations is crucial to minimise these errors and ensure that dose delivery remains accurate across all targets, especially in complex clinical scenarios. This study aims to evaluate the robustness of SIMT SRS treatment plans against a range of clinically relevant uncertainties by implementing different planning optimisations. Specifically, this study investigates how various optimisation strategies impact plan performance, focusing on the treatment plan’s ability to maintain a high target coverage while also minimising radiation dose to healthy tissues under different error scenarios. The set of original (OG) plans was first generated using standard planning protocols, then a set of three additional planning optimisations, including a monitor unit prioritisation (MU), aperture shape control (ASC), and a combination of the two (MU+ASC), were applied. The complexity metrics of these original plans were evaluated, and it was determined that the MU and MU+ASC optimisation methods reduced the complexity of the plans in comparison to the OG plans. A set of ten simulated errors was then introduced to the plans for each patient and optimisation methods. The robustness of each plan was investigated by evaluating the clinically relevant doses to organs at risk (OARs), PTV100%, GTV D98%, and the brain V12Gy dose.</p
Improving Skeleton-based Action Recognition
No full textSkeleton-based human action recognition is a vibrant research area in computer vision; however, existing approaches often struggle to fully capture the complex spatiotemporal dynamics and discriminative features necessary for achieving robust recognition. This thesis introduces four novel methods that enhance action recognition performance in both supervised and self-supervised paradigms.First, a temporal pooling algorithm, termed Asynchronous Joint-Based Temporal Pooling (AJTP), is proposed, which selectively aggregates informative cross-joint and crosstime dynamics. Unlike conventional pooling methods that treat all joints and frames uniformly, often leading to the aggregation of less discriminative features, AJTP is a leamable pooling algorithm that adaptively focuses on the most discriminative features for action recognition. It integrates seamlessly with both Graph Convolutional Networks (GCN) and Transformer-based models, significantly improving feature aggregation in supervised settings. Second, Spatial-Temporal Joint Density (STJD) is proposed as a novel metric to overcome the limitations of existing self-supervised approaches, which are often based on the assumption that only moving body joints and parts are informative. STJD dynamically quantifies the interactions between moving and static joints, enabling the effective identification of a subset of discriminative joints. STJD can be used in both contrastive and reconstruction-based unsupervised frameworks to guide the learning of discriminative representations. Third, a Unified Generative and Discriminative Learning (UGDL) framework is introduced. Traditional self-supervised methods tend to yield representations that either capture overly coarse sequence-level features or focus excessively on low-level details. In contrast, UGDL jointly optimizes explicit objective functions that balance representational capacity with discriminative ability, thereby producing more robust representations for downstream tasks. Lastly, DiffMotion is proposed to explicitly model uncertainty in skeletal motion dynamics, addressing a key limitation in current self-supervised learning approaches. By integrating denoising diffusion probabilistic techniques into the reconstruction framework, DiffMotion models the motion distribution of masked regions conditioned on visible joints, leading to robust action recognition.Together, these contributions advance the state of the art in skeleton-based human action recognition, both theoretically and practically. They provide significant improvements in both supervised and self-supervised learning frameworks, establishing a new benchmark for future research.</p
Improving the quality and safety of hospital care for older people, focusing on two geriatric syndromes: falls and pain
No full textBackground In recent years, there has been an increase in the use of hospital services by older people. One in nine Australians visiting hospitals experience complications; this increases to one in four for those staying overnight. The increased vulnerability experienced by older people poses considerable challenges in their care, making some of them more susceptible to complications known as geriatric syndromes; for example, falls, delirium, and pain, all of which discernibly impacts the quality and safety of care and increases the risk of adverse outcomes. Falls are a major concern, accounting for 40% of injuries in acute care settings, despite preventive strategies. Similarly, pain is often under-recognised and undertreated, especially in older people with cognitive impairment.The Overall Aim of this PhD This PhD aimed to improve the quality of care and safety of older people in acute care by addressing these two key issues: falls in older people and pain in older people with cognitive impairment. This PhD was conducted in the specific context of a hospital within a local health district in New South Wales (NSW), Australia.Specific Aims of the Two Studies Study 1: This study aimed to explore nurses’ and patients’ perceptions of falls using a reflective model chosen by the nurses and to identify strategies to reduce falls among older people on the participating wards. Study 2: The aim of this study was to improve the assessment and management of pain in older people with cognitive impairment on the participating wards.Methods The overarching methodology of this PhD was action research (AR). Study 1 employed a participatory action research (PAR) approach and involved quantitative and qualitative data collection and analysis. It comprised four phases, and two PAR cycles, with quantitative data on the number of falls and qualitative data gathered through focus groups and reflections from staff and patients. Information from the first phase informed the subsequent phase, fostering collaboration between nurses and patients to develop innovative preventive strategies. These solutions were then implemented and evaluated in conjunction with the participating ward staff.Study 2 used AR and was guided by the Integrated Promoting Action on Research Implementation in Health Services (i-PARIHS) framework, which is characterised by four primary constructs: facilitation, innovation, recipients, and context. Evidence was collected through audits, a knowledge survey, and focus groups conducted in participating wards. The data were analysed and collated according to the specific context of each ward. Facilitation played an instrumental role, an evidence-based educational intervention based on feedback from the nurses, and co-designed solutions were developed to enhance pain assessment and management in older people. These were then implemented and evaluated to enhance pain assessment and management.Findings In Study 1, qualitative findings revealed key themes: "It was out of my control," "Falls could have been prevented," and "It was very sad" (Nurses); and "I thought I will be fine," "No one came for a long time," "I fell because...," "You can help me," and "I was hurt" (Patients). In cycle 1, initial interventions reduced falls rates in all wards, although the effects varied. Ward A showed notable decreases followed by a slight increase, Ward B consistently maintained the lowest rates, and Ward C showed fluctuating but reduced rates. In cycle 2, Ward A and Ward B demonstrated promising initial decreases, with slight increases in post-intervention, while Ward C had the highest and most variable rates. Fall-related injuries were minimal and stable across both cycles, with no deaths reported.In alignment with the i-PARIHS framework, Study 2 underscored the significance of innovation, recipients, context, and facilitation in driving positive change among nurses. Data from audits, knowledge surveys and focus groups spotlighted the pivotal role of facilitation in the success of an educational intervention. Qualitative themes highlighted nurses’ perceptions: “inconsistencies in pain assessment and management”, “pain assessment is too hard”, and “ways to improve the pain assessment and management”. Tailored interventions were implemented successfully, e.g., pain champions, and development of a resource manual. This resulted in a considerable increase in the use of behavioural pain assessment tools and administration of analgesics post-assessment. There were notable increases in the nurses' knowledge and confidence in assessing and managing pain in older people with cognitive impairment.Conclusion This study highlights the importance of collaborative, educational, and context-focused interventions for addressing falls prevention and pain management. The findings suggest leadership and facilitation are paramount for implementing solutions for these geriatric syndromes. These results can inform future research, clinical practice, and policy development to enhance care quality for older patients in hospital.</p
A socio-technical approach to the design and implementation of digital health solutions
No full textThe abstract for this item has not been populated.</p
A digital twin simulation platform for 4D millimeter-wave radar-based navigation in underground mining environment
No full textIn underground mining, safe navigation and quick decision-making are critical and essential, especially in emergencies. Environmental conditions, such as, visibility, high humidity, and high amount of dust suspension, make traditional camera or LIDAR-based navigation risky and unreliable. Radars, particularly 4D millimetre-wave (MMW) radar, are relatively robust against these environmental factors, which make them suited to underground mine environment navigation applications. However, iterative testing and refinement of radar systems in real mines remain costly, time-consuming, and potentially hazardous.To address these challenges, a digital twin simulation platform for developing and evaluating a 4D MMW radar-based navigation system for underground vehicles is presented. In the platform, a complex underground mining environment is reconstructed using Gazebo and ROS2, including two specific structures: flat-roof and arched-roof mine tunnels inspired by Queensland Mine Rescue Service (QMRS) conditions. Various obstacles, including rocks, overhead signs, lights, ladders, and a worker model, are integrated into the arched-roof tunnel to evaluate the radar’s detection capabilities. A simplified driftrunner-based vehicle model was designed with radars mounted at two heights: 1.25 m and 0.3 m. Radar parameters were set according to ARS548 hardware specifications using Gazebo’s plugin, while post-processing introduced Gaussian noise and random point cloud reduction to simulate real-world data sparsity and uncertainty.Simulation tests were conducted to analyse radar performance at three distances (5 m, 10 m, and 15 m) along the tunnel centre and two radar heights. Results indicate that radar mounting height significantly influences detection capabilities. At a height of 1.25 m, elevated features, such as, overhead signs and lights can be captured effectively, but ground-level obstacles remained undetected at close range. In contrast, at a lower radar height of 0.3 m, ground-level obstacles can be detected with greater clarity, but elevated features were less distinguishable. This highlight is a tradeoff between the vertical coverage and the ground-level detection, showing the requirement for optimized radar positioning for improved navigation performance.The developed platform allows safe, iterative testing of radar-based systems under simulated underground conditions, providing key insights into sensor configuration, obstacle detection, and point cloud processing. Although currently focused on simulation, the platform supports future integration with physical radar systems to further refine radar plugins and algorithms. This work ultimately aims to enhance safety and decision-making in underground operations.</p
Chloride penetration of shotcrete with crushed waste glass aggregates
No full textShotcrete is a type of concrete sprayed onto surfaces using a nozzle, making it ideal for applications with irregularly shaped substrates such as rock support in deep mines and retaining walls. Unlike conventional concrete, however, shotcrete requires a mix design with enhanced pumpability which is typically achieved by increasing sand content. On the other hand, the surge in construction projects has led to sand depletion due to excessive natural sand extraction, resulting in environmental damage, habitat loss, and beach erosion. To mitigate the sand depletion issues, alternative synthetic aggregates made from crushed waste glass (CWG) have been explored in recent years as substitutes for natural sand in shotcrete production. Evaluating CWG's feasibility in shotcrete involves assessing its performance to ensure it meets standard requirements, particularly its longterm durability. One critical durability factor is shotcrete’s resistance to chloride penetration. Chloride ions, commonly originating from seawater or contaminated groundwater, can infiltrate shotcrete, corrode embedded rebar, and compromise structural integrity, ultimately reducing its lifespan. This study tested shotcrete mixes with varying percentages of CWG for chloride penetration resistance using the Rapid Chloride Penetration Test (RCPT) and the bulk resistivity method. Results demonstrated that incorporating CWG in shotcrete significantly improved durability, showcasing its potential as a sustainable alternative in shotcrete production.</p
Visualisation Techniques for Unveiling the Influence of Distributed and Renewable Generation Units in Distribution Networks
No full textThe abstract for this item has not been populated.</p
Shear testing of rock bolts for hard rock mining
No full textThe shear capacity of hollow rock bolts is questionable due to the lack of experimental data. To address this limitation, the shear strength of three different types of hollow bolts, including “X”, “S” and “E” grade rock bolts were studied using the double shear test method under hard rock conditions. The outcomes revealed that the shear force depends on the bolt type. The “X” grade hollow bolt can tolerate larger vertical movement and shear force. In addition, it was found that the shear load capacity of hollow bolts is not necessarily around 70% of UTS. The results showed that the “X” and “S” grade bolts are more brittle due to their higher UTS and lower elongation capacity.</p
Trust: Four Studies on Innovation, Risk, and Market Dynamics in Decentralized Finance
No full textThis thesis is comprised of four studies on the role of trust in financial transactions, from 5,000 B.C. to contemporary cryptocurrency markets.The first study (Chapter 3) evaluates the literature on the evolution of financial trust, tying it to modern advances in financial technology and decentralized finance. Most crucially, this chapter recounts the pre-modern use of ‘virtual’ currency, including via promises and orally traded stones. It also examines early proto-stablecoins, reinforcing how many of today’s financial needs–and early versions of contemporary solutions–existed in the past. It also demonstrates the extreme importance trust plays in financial transactions, and the many, many ways that trust is sought and demonstrated. From Roman Empire trade receipts created in ancient triplicate, to paper money in 7th Century China, as well as innovations which have long since vanished, financial innovation has long been an important part of every economy of note–and every successful innovation must be trusted to survive.The following three studies examine contemporary cryptocurrency markets to further our understanding of trust at the frontier of contemporary finance.The first of these cryptocurrency studies (Chapter 4) analyzes perpetual futures, a novel derivatives contract first proposed by Nobel Laureate Robert Shiller, but only fully implemented in cryptocurrency markets. In addition to looking at spillover effects between perpetual and traditional financial markets, this chapter examines the unique trust structure of perpetual futures–which on Binance includes socialized losses and a reserve fund denominated in cryptocurrencies–as well as 'funding rates', periodic payments between traders. It also tests arbitrage opportunities in Bitcoin quarterly futures, contributing to our understanding of market efficiency in cryptocurrency markets over time.The next study (Chapter 5) examines the collapse of the TERRA/LUNA algorithmic stablecoin pair, analyzing contagion effects spread across markets, contributing to our understanding of how distrust spreads across cryptocurrency markets, while also testing trust in stablecoin issuers via market prices of issued assets. This chapter finds that traders have clear preferences for clearly defined, transparently-backed assets in a crisis. This is demonstrated by the market price of Tether, which traded at a 5% discount, while other stablecoins with more transparent and regulated reserve structures traded at a premium. Given Tether is supposed to trade for essentially $1 at all times, any non-minor deviation from the market price offers arbitrage profits to speculators, assuming they have faith in the underlying asset. That Tether traded at a discount for so long, while rivals traded at a premium, opens a window into trader behavior and the value of trusted and transparent reserve structures in a crisis.The final study (Chapter 6) investigates the collapse of FTX, analyzing how contagion effects spread across markets during an exchange collapse. This study examines the price action of major cryptocurrencies stranded on the FTX exchange during a unique period where trading was allowed but withdrawals were almost entirely halted. This period provides insight into trader behavior and asset preference in advance of exchange bankruptcy. Major divergences between prices on FTX and the wider cryptocurrency markets were found, demonstrating how the unique pressure of an exchange collapse and withdrawal halt can dramatically influence market prices. The study examines an on-chain FTX-controlled Ethereum wallet, providing further detail about the FTX collapse while partially confirming FTX statements around the withdrawal halt. Verifying these statements, to the extent possible, was important as FTX was accused of fraud.As a whole, this dissertation enhances our understanding of financial trust in contemporary digital markets, with empirical evidence demonstrating how contemporary financial challenges–from counterparty risk, identity verification, asset backing structures, market contagion, liquidity crises, and reputational risks–are enduring aspects of finance across eras. This research provides frameworks for evaluating market trust in novel financial asset classes while placing the contemporary decentralized financial space within the broader evolution of financial markets, highlighting the ways financial innovation can both create and destroy trust in value exchange.</p
Gardens, agency and citizenship of people with dementia: An opportunity to dig deeper?
No full textThe abstract for this item has not been populated.</p