143174 research outputs found
Sort by
Deep learning for cortical surface reconstruction
No full textCortical surface reconstruction facilitates both visualisation and quantification of the cerebral cortex, playing a pivotal role in the diagnosis of neurological disorders and the characterisation of cortical folding patterns. Due to the highly folded anatomical structure and intrinsic topological constraints, it is challenging to extract anatomically and topologically accurate cortical surfaces from brain magnetic resonance imaging (MRI) data. In this thesis, we leverage advanced geometric deep learning (DL) techniques and present a series of DL-based frameworks for fast and explicit cortical surface reconstruction from adult, fetal and neonatal brain MRI scans.
Firstly, we propose a novel deep neural network architecture for explicit and topology-preserving cortical surface reconstruction end-to-end from adult brain MRI. To prevent surface self-intersections, we further introduce neural ordinary differential equations to learn diffeomorphic surface deformations for cortical surface reconstruction. Secondly, we develop customised DL-based frameworks for fetal and neonatal subjects that undergo rapid brain development. To tackle considerable brain variations across different ages, we utilise an attention mechanism to learn neonatal cortical surface construction conditioned on the ages of neonates. Furthermore, we devise a weakly supervised framework for fetal cortical surface reconstruction supervised by brain segmentations, thereby eliminating the reliance on pseudo ground truth cortical surfaces generated by traditional neuroimage processing pipelines. Finally, we present a fast and robust DL-based pipeline for cortical surface-based structural MRI processing of developing human brains.
We demonstrate that the DL-based cortical surface reconstruction approaches proposed in this thesis achieve superior geometrical and topological accuracy, fast inference within only a few seconds, and high adaptability to brain MRI acquired from subjects in various age groups including adults, fetuses and neonates.Open Acces
Integrated numerical modeling and data-driven techniques for thermal effluent simulation in coastal waters
No full textThis thesis investigates the integration of numerical modelling and data-driven techniques to improve the simulation of thermal effluent discharged from coastal power plants into tidal waters. Uncertainties in key effluent characteristics, particularly the discharge flow rate and excess temperature relative to ambient water, pose significant challenges to accurate model representation, often resulting in errors in seawater temperature predictions. These challenges are further complicated by the dynamic nature of tidal systems, where strong advective and dispersive processes govern the transport and spread of discharged thermal plumes.
Three methodologies were integrated into the modelling framework to estimate discharge parameters and improve the accuracy of seawater temperature simulations. These methodologies include: (1) data assimilation (DA) using the Ensemble Kalman Filter (EnKF) for state estimation of seawater temperature; (2) Bayesian optimisation to calibrate hydrodynamic model parameters using temperature observations from in-situ profiles, thermal unmanned aerial vehicle (UAV) imagery, and satellite data; and (3) Large-Scale Particle Image Velocimetry (LSPIV) to integrate discharge rate estimates from optical UAV footage into the hydrodynamic model. Each approach was systematically evaluated for its effectiveness in reducing parameter uncertainties and improving model accuracy.
The data assimilation findings demonstrate that optimising data collection locations and their assimilation frequencies significantly reduces errors in simulated seawater temperature by maximising the spread and retention of DA adjustments across the modelled domain. Bayesian optimisation highlighted the critical role of spatial coverage in temperature observations on parameter estimation outcomes, demonstrating the importance of strategic data collection. LSPIV-derived discharge offers a viable, cost-effective method for estimating outflows from coastal power plants in tidal environments, with its integration into hydrodynamic models enhancing simulations and complementing calibration efforts.Open Acces
Origami based deployable surfaces: an optimization approach
No full textThis thesis explores using origami and optimisation approaches for designing deployable structures. In particular, the problem of solid-surface deployable reflectors is studied, and optimal designs are pursued under simultaneous constraints including rigid foldability, double curvature and complex collision between finite thickness components.
The research includes three aspects. First, optimality of origami designs is pursued. Many origami designs as well as deployable reflector designs are defined and presented without discussion of whether they are the best possible design. Here, an optimisation approach is used to produce best results for predefined design problems with specific requirements. Second, a formalised design methodology is pursued. A process is developed where a deployable structure design problem is translated into an origami design problem, and the origami design problem into an optimisation problem. This provides a formalised and mostly automated path towards optimal deployable structures designs. Third, design results for the class of stringent design problems studied here are explored. Complex local constraints such as collision are solved via adjustment of the global folding kinematics. It is shown that using the novel design approach developed here, optimal results with tangible improvement upon previous designs, as well as unforeseen features can be obtained.
The two main foci of the work included in this thesis are two deployable reflector concepts based on the flasher and Miura-ori pattern respectively. The former introduces a new variant of the flasher pattern involving cut creases. Advances include achievement of full rigid foldability under double curvature, optimised compact stowage, and details such as elimination of gaps between panels, all of which are rarely achieved by similar designs. The latter concept uses the Miura-ori pattern and the Hoberman linkage to construct a novel kinematic architecture which has fully adjustable geometry for both the deployed and stowed states, whilst simultaneously achieving single-degree-of-freedom folding.Open Acces
Aviation sector decarbonisation as a case of deep uncertainty: the need for an integrative, exploratory, and interdisciplinary approach
No full textGlobal demand for aviation is expected to double by 2050. This is set against the need to cut aviation CO2 emissions from 1 GtCO2 pa today to net zero over the same period. Using the UK aviation market as a proxy for the global market, we apply a Robust Decision Making (RDM) to develop integrative insights within a single analytical paradigm than discrete orthodox decision support analysis.
This approach is justified based on a critical examination of the sector characteristics and the divergent short-term motivations of aviation actors as a case of deep uncertainty. RDM explicitly embraces deep uncertainty across a number of metrics which allows multiple values and diversity among stakeholders and viewpoints, and in which modelling can exist in an iterative exchange with policy development rather than separate from it. This approach has particularly highlighted the critical significance of asset stranding, the oligopolistic structure in aerospace manufacturers and fuel suppliers, alongside the monopsonies in airlines, as current barriers to progress within a single integrative analytical paradigm.
This contribution highlights the need for the application of exploratory and interdisciplinary approaches to aviation decarbonisation transitions analysis to better inform aviation sector net zero strategies. It can improve the interdisciplinarity of analysis across technology, policy, and finance, and explore the extent of uncertainty so that robust strategies can be designed as well as generate insight into systematic upstream requirements
Advancing embodied virtual reality for motor learning and neurorehabilitation
No full textAdvancements in Embodied Virtual Reality (EVR) can revolutionise motor learning and rehabilitation by offering immersive and ecologically valid environments. This study refines an EVR setup to investigate motor learning using a realistic yet controlled billiards task.
The setup integrates physical interactions with virtual experiences, allowing participants to use real-world objects, including a cue stick and balls. Enhancements include a redesigned pool table to control task difficulty, refined virtual ball velocity profiles for more realistic dynamics, and novel feedback mechanisms that isolate error-based and reward-based learning.
Pilot studies with naive and moderately experienced pool
players demonstrated increased realism, engagement, and experimental control. The system supports the study of adaptive skill acquisition in naturalistic contexts, addressing the challenges of balancing experimental control with real-world applicability.
The improved EVR setup provides a powerful tool for
studying motor learning and translating laboratory insights
into real-world neurorehabilitation applications, supporting
adaptive skill acquisition in naturalistic settings.
Clinical relevance — The motor learning insights from the
EVR setup hold significant implications for neurorehabilitation, providing a robust platform to design adaptive and engaging rehabilitation programs for conditions like stroke recovery and motor impairments, fostering skill relearning in controlled yet naturalistic settings
Investigating behavioural correlates of sleep states in drosophila
No full textSleep is a universal yet diverse phenomenon across the animal kingdom. Despite its evolutionary conservation, sleep varies significantly among mammals and even more so when compared to birds, reptiles, fish, cephalopods, and insects. Recent studies have bridged the understanding of sleep across different animal classes, revealing that multiphasic sleep, characterised by distinct stages like REM and NREM in mammals, may be present across a wider range of species, though not necessarily in a structurally or functionally equivalent manner. To generate a general theory of sleep, it is essential to accurately identify and classify sleep types across species to facilitate meaningful comparisons.
In the past two decades, Drosophila melanogaster has emerged as a leading model for studying sleep. However, most research has treated Drosophila sleep as monophasic, typically defined by the cessation of movement for five minutes or more. These metrics have become outdated, as recent studies have uncovered evidence of multiphasic sleep patterns in Drosophila, including brain activity indicative of different sleep phases, behavioural markers of deeper sleep, and shorter sleep latencies. To understand sleep in Drosophila, it need to integrate these new findings into our analysis toolbox without compromising the high-throughput nature of Drosophila research.
This study investigates behavioural markers of sleep, utilising both high-throughput, low-resolution machine vision tracking data and individual, high-resolution limb tracking to identify behavioural correlates of sleep states. Our findings reveal that low-resolution data cannot currently classify sleep states or depth more accurately than the traditional five-minute inactivity rule. However, it does allow for more detailed analysis of sleep patterns, particularly when characterising sleep in new Drosophila species or mutants. High-resolution tracking, while capturing sleep postures, has limited applicability for standard sleep research due to experimental setup constraints. This work aims to be a foundation for developing more refined metrics for understanding sleep in Drosophila.Open Acces
Analysis of airway inflammation demonstrates a mechanism for T2-biologic failure in asthma
No full textBackground
Targeted type 2 (T2) biologics have transformed asthma care, but the clinical response to biologic therapy varies between patients.
Objective
We sought to assess airways inflammation in T2-high asthmatic patients treated with anti–IL-5 biologics to investigate whether differential mechanism of airway inflammation explains varied response to biologics.
Methods
Proteomic analysis (Olink, 1463 protein panel) and high-sensitivity cytokine analysis (ELISAs) were performed on induced sputum from T2-high severe asthmatic patients in the UK multicenter Mepolizumab EXacerbation study. Samples included were pre-mepolizumab (n = 28), stable on mepolizumab (n = 43), and at first exacerbation (n = 26).
Results
Clustering of sputum proteins while stable on mepolizumab identified 2 clusters. Cluster 1 had increased differentially expressed sputum proteins pre-mepolizumab, stable on mepolizumab, and at exacerbation. Patients in cluster 1 were younger at diagnosis, had a longer duration of asthma, lower FEV1%, and higher 5-Question Asthma Control Questionnaire score on mepolizumab. Cluster 1 had increased expression of proinflammatory cytokines (IL-1β, IL-6, and soluble IL-6R), epithelial alarmins (thymic stromal lymphopoietin [TSLP] and IL-33), and neutrophil activation (myeloperoxidase [MPO], neutrophil elastase [NE], and neutrophil extracellular trap concentration [NET]). All patients were T2-high with no difference in fractional exhaled nitric oxide, eosinophil number, or activity (eosinophil-derived neurotoxin, EDN) across the 2 clusters.
Conclusions
In a cohort of T2-high severe asthmatic patients, a subgroup of patients with long duration of disease had worse clinical parameters, increased sputum proteins with increased markers of neutrophil activity, proinflammatory cytokines, and epithelial alarmins even when stable on mepolizumab. This suggests the presence of biology not treated by targeted T2 biologics, which may contribute to poorer outcomes on biologics and could be a treatable airways trait in severe asthma
Roux-en-Y gastric bypass, adjustable gastric banding, or sleeve gastrectomy for severe obesity (By-Band-Sleeve): a multicentre, open label, three-group, randomised controlled trial
No full textBackground
The health risks of severe obesity can be reduced with metabolic and bariatric surgery, but it is uncertain which operation is most effective or cost-effective. We aimed to compare Roux-en-Y gastric bypass, adjustable gastric banding, and sleeve gastrectomy in patients with severe obesity.
Methods
By-Band-Sleeve is a pragmatic, multi-centre, open-label, randomised controlled trial conducted in 12 hospitals in the UK. Eligible participants were adults (aged ≥18 years) meeting national criteria for metabolic and bariatric surgery. Initially, a 2-group trial (Roux-en-Y gastric bypass versus adjustable gastric banding) became a 3-group trial to include sleeve gastrectomy at 2·6 years from study opening, when it became widely used in the UK. Co-primary endpoints were weight (proportion achieving ≥50% excess weight loss) and quality-of-life (EQ-5D utility score) at 3 years. If the proportion achieving at least 50% excess weight loss was non-inferior (<12% difference between groups) and quality-of-life was superior, sleeve gastrectomy and Roux-en-Y gastric bypass were considered more effective than adjustable gastric banding, and sleeve gastrectomy more effective than Roux-en-Y gastric bypass. Cost-effectiveness of the procedures was compared. This trial is registered with ClinicalTrials.gov, NCT02841527, and ISRCTN, 00786323.
Results
Between Jan 16, 2013, and Sept 27, 2019, 1351 participants were randomly assigned; five withdrew consent and 1346 (mean age 47·3 [SD 10·6] years, 1020 [76%] women, 324 (24%) men, and two with missing data, mean weight of 129·7 kg [23·6] and mean BMI of 46·4 [6·9] kg/m2) were included in this report. Of 1346 participants, 462 (34%) were in the Roux-en-Y gastric bypass group, 464 (34%) in the adjustable gastric banding group, and 420 (31%) in the sleeve gastrectomy group. 1183 (88%) participants underwent surgery. 276 (68%) of 405 participants in the Roux-en-Y gastric bypass group, 97 (25%) of 383 participants in the adjustable gastric banding group and 141 (41%) of 342 participants in the sleeve gastrectomy group achieved at least 50% excess weight loss (adjusted risk difference: Roux-en-Y gastric bypass vs adjustable gastric banding 41% [98% CI 34 to 48]; sleeve gastrectomy vs adjustable gastric banding 15% [5 to 24]; sleeve gastrectomy vs Roux-en-Y gastric bypass, –26% [–36 to –16%]). Mean EQ-5D scores were 0·72 for Roux-en-Y gastric bypass, 0·62 for adjustable gastric banding, and 0·68 for sleeve gastrectomy (adjusted mean difference: Roux-en-Y gastric bypass vs adjustable gastric banding 0·08 [0·04 to 0·12], sleeve gastrectomy vs adjustable gastric banding 0·05 [0·01 to 0·09], and sleeve gastrectomy vs Roux-en-Y gastric bypass –0·03 [–0·07 to 0·01]). 1651 adverse events were reported following surgery (5·7 per year after sleeve gastrectomy, 6·0 per year after Roux-en-Y gastric bypass, and 4·6 per year after adjustable gastric banding). There were 11 deaths from randomisation to 3 years: one attributable to surgery (in the adjustable gastric bypass group, during the surgical admission) and ten not attributable to surgery (four each in the Roux-en-Y gastric bypass and adjustable gastric banding groups and two in the sleeve gastrectomy group). Roux-en-Y gastric bypass was most cost-effective.
Interpretation
Roux-en-Y gastric bypass and sleeve gastrectomy are more effective than adjustable gastric banding. Sleeve gastrectomy has inferior weight loss and lower mean quality of life score compared with Roux-en-Y gastric bypass. Based on this evidence, it is recommended that patients electing to have metabolic and bariatric surgery are advised to have Roux-en-Y gastric bypass. Where contraindicated or unfeasible, sleeve gastrectomy should be offered. This evidence does not support adjustable gastric band as standard treatment for severe obesity.
Funding
National Institute for Health and Care Research Health Technology Assessment Programme
Simultaneous monitoring of tyrosinase and ATP in thick brain tissues using a single two‐photon fluorescent probe
No full textCellular redox homeostasis and energy metabolism in the central nervous system are associated with neurodegenerative diseases. However, their real-time and concurrent monitoring in thick tissues remains challenging. Herein, a single dual-emission two-photon fluorescent probe (named DST) is designed for the simultaneous tracking of tyrosinase (TYR) and adenosine triphosphate (ATP), thereby enabling the real-time monitoring of both neurocellular redox homeostasis and energy metabolism in brain tissue. The developed DST probe exhibits excellent sensitivity and selectivity toward TYR and ATP, with distinctive responses in the blue and red fluorescence channels being observed without spectra crosstalk. Using this probe, the correlation and regulatory mechanism between TYR and ATP during oxidative stress are uncovered. Additionally, the two-photon nature of this probe allows alterations in the TYR and ATP levels to be monitored across different brain regions in an Alzheimer's disease (AD) mouse model. Notably, a significant decrease in ATP levels is revealed within the somatosensory cortex (S1BF) and caudate putamen brain regions of an AD mouse, alongside an increase in TYR levels within the S1BF and laterodorsal thalamic nucleus brain regions. These findings indicate the potential of applying the spatially resolved regulation of neurocellular redox homeostasis and energy metabolism to treat neurodegenerative diseases
Magnetic field effects on the corrosion behavior of magnetocaloric alloys LaFe13.9Si1.4Hy under ferromagnetic states
No full textLa(Fe,Si)13-based alloys, with giant magnetocaloric effect, still encounter significant degradation issues prior to commercial viability. In this work, the corrosion behavior of ferromagnetic La(Fe,Si)13Hy was investigated with electrochemical linear polarization resistance measurements under conditions with zero, 1 T parallel, and perpendicular magnetic fields, mimicking practical application scenarios. The results demonstrated that both parallel and perpendicular magnetic fields had a suppressive effect on corrosion rates due to the combined influence of magnetohydrodynamic forces and magnetic field gradient forces. The inhibiting efficiency of the parallel field decreased with increasing exposure period, while that of the perpendicular field continued to increase over time. The magnetic field also affected the relative proportion of rust phases, and thereby the protectiveness of the rust layer. This highlights the importance of conducting experiments under service conditions to understand the degradation mechanisms of magnetic cooling devices