213476 research outputs found
Sort by
Straight Time and Queer Utopia: Anthony Bridgerton and Kate Sharma's Disorientating Desire in the Regency
No full textThis article analyses the romance between Anthony Bridgerton (portrayed by Jonathan Bailey) and Kate Sharma (Simone Ashley) in season two of the Shondaland production of Bridgerton. In dialogue with queer theory, principally the notions of straight time and queer utopia by José Esteban Muñoz, reproductive futurity by Lee Edelman and queer orientation by Sara Ahmed, I argue that Kate and Anthony’s relationship can be understood as “queer” within the Regency-inspired diegetic context of the series. I examine this context, as well as the historical details upon which it is based, to underscore the marriage expectations that Anthony seeks to conform to, but which his desire for the queer Kate veers him away from. Through close reading, I argue that this queer dynamic is visually reinforced, particularly through their doubling and the same-sex coding of their relationship. Ultimately, I show that instead of remaining “unhappy queers,” in line with Ahmed’s scheme, they lose their queer status in the resolution of the series as they are reincorporated into the heteronormative dominant culture through their “happy ending.” At stake, I argue, in the initial framing of their relationship as queer, is the heightened audience investment through prohibition and delayed gratification in an era of relative permissiveness
Evaluation of Uncertainty-Aware Multi-software Ensembles for Hippocampal Segmentation
No full textAccurate hippocampal segmentation can be a useful tool for diagnosing and monitoring neurological conditions such as Alzheimer’s disease and epilepsy. While numerous automated segmentation methods exist, their clinical adoption remains limited. Reliable uncertainty assessment can enhance trust and facilitate clinical translation. This study evaluates five heterogeneous hippocampal segmentation methods InnerEye, ASHS, FastSurfer, HippoSeg, and FreeSurfer—across two dementia datasets and one epilepsy dataset. The sub-ensemble containing InnerEye, FastSurfer, and HippoSeg emerged as both accurate and efficient, highlighting the feasibility of balancing computational cost and performance. Additionally, ensemble-derived uncertainty quantification with sample variance, mutual information, and predictive entropy is shown to reduce inaccurate segmentations by flagging low-confidence cases, potentially providing a mechanism for automatically escalating ambiguous cases for expert assessment
Advanced Preclinical Cardiac MRI of Heart Diseases
No full textCardiovascular diseases remain a leading cause of morbidity and mortality, with more than three million people are estimated to suffer from heart disease worldwide [1]. Magnetic resonance imaging (MRI) has grown in importance as a non-invasive method of evaluating cardiac tissue. T1 and T2* mapping in MRI allows for the quantitative characterisation of tissue composition and function, which aids in the diagnosis and treatment of many cardiac diseases. The aim of this study was to develop and validate the alternative quantitative cardiac MRI techniques for non-invasive characterisation of myocardial tissue properties and oxygenation, as well as to examine their response to pharmacological stress with dobutamine. To accomplish this objective, the project was divided into four main parts. The first part focuses on evaluating and validating the variable flip angle (VFA) T1 sequence with different hardware settings using phantom and in vivo on the 9.4T Bruker MRI. Then, in the second part, the validated VFA T1 sequence was used with MRI contrast agent manganese, a calcium analogue, to evaluate its usability for assessing myocardial viability in normal, remote, and infarcted myocardium. In the third part, a ratiometric T2* sequence was developed with the aim of gaining insight into tissue oxygen dynamics and myocardial health. This sequence was then used to assess myocardial oxygenation with different gas mixtures in the hypertensive heart. The final section extends the research focus to introduce novel dobutamine-stress T1-Manganese Enhances MRI (T1-MEMRI) and ratiometric T2*-MRI to assess Ca2+ uptake and myocardial oxygenation, respectively. Collectively, this thesis highlights how novel quantitative MRI techniques can be utilised to generate detailed assessment of myocardial function, viability and oxygenation under a variety of normal and pathological conditions
Parsimonious Time Series Modelling of High-dimensional Data with Linear and Non-Linear Models
No full textTime series models perform statistical analysis and predictions based on sequential data, with two trends emerging in their development: high dimensionality and non-linearity. The first trend results from advances in data collection and storage capabilities, and the second arises from the dynamic evolution of sequential data that challenges traditional static linear models. In this thesis, we study these two trends in Bayesian Vector Autoregression (VAR), chosen for its flexibility and widespread applications across fields. The Bayesian framework enables the incorporation of well-designed priors for unknown parameters while providing insight into parameter uncertainty. The first part of the thesis focuses on addressing the proliferation of parameters, a challenge inherent to high dimensionality, through tensor VAR, which treats the coefficient matrix as a third-order tensor and conducts tensor decomposition to achieve parsimony. We apply different Bayesian techniques to induce shrinkage and achieve convergent Markov chains. In the second part, we introduce time-varying parameterization to the tensor VAR to capture the evolving interrelation among time series, enabling the model to accommodate both trends. The final part explores factor augmented VAR (FAVAR), an alternative VAR framework that reduces dimensionality by extracting factors from high-dimensional data. We develop a novel approach which combines time-varying parameterization in the VAR component with a proposed Grouped Sparse autoencoder, alleviating identifiability and interpretability issues of the standard autoencoder. To demonstrate the merits of the models proposed in these three parts, we apply them to macroeconomic data and functional magnetic resonance imaging data
Temporally Stable Monocular Depth Estimation in Surgical Vision
No full textRecent foundational models have unlocked numerous possibilities for computer-assisted interventions. A critical advancement in this field is the precise estimation of dense relative depth on surgical videos, essential for understanding the 3D positioning of surgical instruments and measuring anatomical structures. However, existing methods often struggle to estimate depth maps that are coherent and smooth over time, leading to noisy and temporally inconsistent depth predictions. We propose TAN, a novel Temporal Adapter Network for monocular depth estimation that enhances the foundational model Depth Anything V2 from image-based to temporally aware depth estimation. Specifically, we design a lightweight temporal adapter and integrate it into the decoder to capture temporal features from consecutive frames. Additionally, we introduce a self-supervised temporal regularization loss, utilizing optical flow to enforce stable depth estimation between consecutive frames. Our experiments, conducted on the SCARED and EndoNeRF datasets, two established benchmarks for evaluating depth estimation models in the surgical domain, demonstrate that the proposed TAN improves both temporal consistency and depth accuracy, achieving at least a 14.29% reduction in OPW and 3.6% in RMSE on SCARED, and 6.2% in OPW and 3.26% in RMSE on EndoNeRF compared to state-of-the-art methods, while running at 97 FPS, making it well-suited for real-time surgical applications
Exploring the Role of Post-Earthquake Reuse of Vernacular Heritage in Enhanced Disaster Resilience and Comfort
No full textIncreasing natural hazards and climate change have validated the disaster resilience of built heritage in the vernacular background through its construction knowledge. This research frames this integrated system as Vernacular Building Heritage (VBH). VBH’s components contain architectural form, context layout and wisdom in building formation, installation, evolution and adaptation to new circumstances. The disaster-resistance and reuse of VBH during post-disaster reconstruction (PDR) also indicate its sustainability after strongly shocked by modernisation and industrialisation, which researchers seldom discuss and explore. This research adopts a post-Wenchuan Earthquake (May 12, 2008) rural reconstruction project in Sichuan Province as a case study, proposing a VBH’s values model for local community within disaster contexts. Historical documents and community/carpenter interviews elucidate how VBH values crystallised within vernacular architecture and its knowledge systems. Concurrently, on-site data collection documents VBH's evolutionary trajectory when responding to rural modernisation. This thesis assesses building performance and VBH’s sustainability through analysis of field investigation data. Findings demonstrate that traditional Chuan-Dou structures continue to fulfil local requirements in post-disaster reconstruction contexts, while contributing to the ongoing conservation of building forms and construction know-how. To extend VBH conservation beyond this case study, the author recommends: enhancing physical and social building performance; maintaining traditional contexts as human-nature equilibrium; safeguarding construction knowledge systems; and establishing sustainable community-centred tourism frameworks. This VBH evaluation framework—assessing building performance and heritage significance—offers transferable methodology for future cases and other heritage typologies
altx: A Python package for adaptive law-based transformation in time series classification
No full textWe introduce altx, an open-source Python package for computationally lightweight and transparent time series classification pipelines. The altx package implements the adaptive law-based transformation, a multiscale feature extraction method that maps raw time series to compact tabular feature vectors by pooling class-labeled law responses across windows and scales. The approach extends the linear law-based transformation with a multiscale shifted-window schedule while preserving transparency. The package provides a GPU-capable PyTorch implementation with an estimator-style interface, enabling straightforward integration into modern machine-learning workflows and interoperability with common scientific Python toolkits. We include illustrative examples and summarize representative benchmark results reported in our companion methodological paper
Training tactile sensors to learn force sensing from each other
No full textHumans achieve stable and dexterous object manipulation by coordinating grasp forces across multiple fingers and palms, facilitated by a unified tactile memory system in the somatosensory cortex. This system encodes and stores tactile experiences across skin regions, enabling the flexible reuse and transfer of touch information. Inspired by this biological capability, we present GenForce, the first framework that enables transferable force sensing across diverse tactile sensors in robotic hands. GenForce unifies tactile signals into shared marker representations, analogous to cortical sensory encoding, allowing force prediction models trained on one sensor to be transferred to others without the need for exhaustive force data collection. We demonstrate that GenForce generalizes across both homogeneous sensors with varying configurations and heterogeneous sensors with distinct sensing modalities and material properties. This transferable force sensing capability is also demonstrated in robot manipulation tasks including daily-object grasping, slip detection and compensation with multi-sensor force coordination. Our results highlight a scalable paradigm for cross-sensor robotic tactile sensing, offering new pathways toward adaptable and tactile memory-driven robot manipulation in unstructured environments
Optical Study on the Impact of Molecular Structure in Potential Biomass-Derived Fuels on SI Engine Performance
No full textIn recent decades, climate change has become a critical issue, with the release of greenhouse gases (GHG) from the combustion of fossil fuels significantly increasing due to the rising demand for vehicles. It is essential therefore to develop alternative energy sources for transport to reduce GHG emissions. The utilisation of fuels from biological sources is one potential solution that can help mitigate climate change. While some biofuels, such as ethanol and biodiesel, are already widely used and contribute to the displacement of fossil fuels. However, the production of fuels from lignocellulosic feedstocks, for example waste biomass, can result in potential fuel molecules of varying molecular structure and it is important to understand how these features impact engine performance and pollutant formation. In this research, four cyclic hydrocarbons with potential as drop-in biomass derived fuels for spark ignition engines, cyclopentane, cyclohexane, cyclopentanone, and cyclohexanone, were selected to investigate the effects of cyclic ring size and the presence of ketone groups on spray penetration and soot luminosity in a gasoline direct injection (GDI) engine. These potential fuel molecules can be derived from lignocellulosic biomass via conversion to xylose and phenol with subsequent hydrogenation and hydrodeoxygenation (HDO), respectively. The molecules were tested both as single fuels and as blends with a fossil gasoline in an optically accessible 2-litre GDI engine with a windowed piston. A high-speed camera and laser illumination were employed to capture spray penetration and soot luminosity. All experiments were conducted at constant RPM and air-fuel ratio (AFR) under lean stratified combustion conditions. Trends were apparent in soot luminosity depending on ring size and the presence of the ketone group. Molecules with a 5-membered ring exhibited reduced soot formation compared to those with a 6-membered ring. Additionally, the spray penetration rate was found to be affected by molecular structural differences, particularly due to density. A higher penetration rate was observed in molecules with a 6-membered ring compared to those with a 5-membered ring, owing to their higher density