143174 research outputs found
Sort by
Isolated hypoglossal nerve mononeuropathy associated with preeclampsia in late pregnancy: a case report
No full textBackground: Neurological complications in preeclampsia are uncommon and typically central in origin. Cranial mononeuropathies, especially involving the hypoglossal nerve, are exceedingly rare.
Case Presentation: We report a case of isolated hypoglossal nerve mononeuropathy in a 31-year-old primigravida at 37 weeks of gestation with preeclampsia. The patient presented with mild dysphagia and dysarthria. Neurological examination revealed tongue deviation to the right, suggestive of left hypoglossal nerve mononeuropathy. MRI brain with T1/T2, FLAIR and DWI/ADC sequences and MR angiography, followed by contrast enhanced MRI of the neck and oral cavity, showed no infarct, haemorrhage, mass lesion or compressive pathology. The patient was managed conservatively with blood pressure optimisation, and symptoms resolved completely within 90 days postpartum.
Conclusion: At least four cases of hypoglossal nerve mononeuropathy associated with preeclampsia have been reported in the literature. This case adds to the emerging evidence that cranial mononeuropathies may represent a rare but important neurological manifestation of preeclampsia and highlights the need for urgent neuroimaging and careful blood pressure control in pregnant and/or postpartum patients presenting with bulbar symptoms
A multimodal slice discovery framework for systematic failure detection and explanation in medical image classification
No full textDespite advances in machine learning-based medical image classifiers, the safety and reliability of these systems remain major concerns in practical settings. Existing auditing approaches mainly rely on unimodal features or metadata-based subgroup analyses, which are limited in interpretability and often fail to capture hidden systematic failures. To address these limitations, we introduce the first automated auditing framework that extends slice discovery methods to multi-modal representations specifically for medical applications. Comprehensive experiments were conducted under common failure scenarios using the MIMIC-CXR-JPG dataset, demonstrating the framework’s strong capability in both failure discovery and explanation generation. Our results also show that multimodal information generally allows more comprehensive and effective auditing of classifiers, while unimodal variants beyond image-only inputs exhibit strong potential in scenarios where resources are constrained
Improving recognition of rickettsial infections in Sri Lanka: the value of combined qPCR and ELISA in acute febrile illness
No full textIntroduction
In Sri Lanka, reliable diagnostics for rickettsial infections are limited. We assessed the burden, seroprevalence, and molecular diversity of rickettsial infections in hospitalized patients with acute undifferentiated febrile illness(AUFI).
Methods
AUFI patients were enrolled from the Western (n=540) and Central (n=260) Provinces. Clinical and exposure data, acute and convalescent sera, and buffy coat/eschar samples were collected. Orientia tsutsugamushi(OT) and Rickettsia spp. were detected using OT-specific (47 kDa) and pan-Rickettsia (17 kDa) qPCR assays. Group-specific IgG ELISAs were performed for scrub typhus(STG), typhus group(TG), and spotted fever group(SFG). Acute infection was defined by qPCR positivity and/or a ≥4-fold rise in IgG titres. Genetic diversity was assessed using OT 56kDa TSA and Rickettsia ompB gene sequencing.
Results
qPCR identified 38/800(5%) STG and 25/800(3%) TG/SFG infections. Among participants with paired sera (n=493), acute infections included 25(5%) STG, 11(2%) TG, and 17(3%) SFG cases. Overall, rickettsioses accounted for 86/800(11%) of AUFI cases. SFG seroprevalence was higher in the Central Province (17% vs 6%), while STG seroprevalence was higher in the Western Province (20% vs 13%). Genetic analysis on OT cases showed clustering with Karp, Kato, Buie, TH1811, TH1826 and Ikeda strains. Two cases of rickettsial infections were speciated as R. felis and R. sibirica. Only 19% of acute cases were clinically recognized and 58% received doxycycline.
Conclusion
Rickettsioses were under-recognized, highlighting the need for combining qPCR and ELISA diagnostics to strengthen clinical management and surveillance
A pilot study to assess the challenges and efficacy of two hearing loss simulations
No full textDeveloping accurate and customisable hearing loss (HL) simulations is crucial for understanding and raising awareness of the challenge faced by individuals with HL. This pilot study assesses challenges in perceptually validating two real-time audio effects plugin HL simulations: the 3D Tune-In (3DTI) Toolkit and the Queen Mary University of London (QMUL) plugin. Both simulate
common HL deficits, with 3DTI offering greater customization. A pilot listening study was conducted involving normal-hearing listeners with simulated HL and those with real HL, focusing on mild-to-moderate high-frequency hearing loss. Audiometric tests and psychoacoustic tasks were employed, including gap and tone detection in noise, perceived sound intensity, and
intelligibility tests. Results from two real HL listeners guided simulation adjustments for normal-hearing participants. Initial findings suggest reasonable accuracy in replicating spectral resolution and perceived sound intensity, but variability in intelligibility and temporal resolution tests indicates room for improvement in both implementations. This study highlights the need for
enhanced customisation to improve accuracy and applicability, offering insights into development challenges. Furthermore, the employed methodology proves to be effective, offering valuable insight into challenges and biases that can occur during testing sessions, while highlighting the necessity for further research. This could include additional HL listeners in order to refine and develop more precise tools for understanding and addressing HL
La Niña, Climate change, high exposure and vulnerability combined led to devastating floods in parts of Southern Africa
No full textCite as Pinto, I. et al., (2026): La Niña, Climate change, high exposure and vulnerability combined led to devastating floods in parts of Southern Africa (WWA scientific report No. 81) World Weather Attributio
Honeybee-mimicking microenvironment-responsive theranostic nanovesicles for enhanced magnetic resonance imaging and targeted thrombolytic therapy
No full textDeveloping a theranostic system integrating specific thrombus targeting, microenvironment-responsive drug release and imaging remains challenging. Inspired by “honeybees finding nectars”, we first designed a honeybee-mimicking nanovesicle (denoted as tPA@THPTN), which mimics honeybees' ability to seek (target), to respond (release drug in the target microenvironment), and to report (provide real-time location feedback) for targeted delivery of tissue plasminogen activator (tPA) to enhance thrombolytic therapy. This biomimetic nanovesicle was constructed through simple self-assembly with thioketal linker-containing lipids for microenvironment response, decorated with fibrinogen-derived peptides for targeting, and surface-coated with 2,2,6,6-tetramethylpiperidin-1-oxyl for enhanced magnetic resonance (MR) contrast. In vitro static and microfluidic studies confirmed the specific binding to activated platelets and efficient thrombus targeting under a flow condition. Efficient thrombus microenvironment response and triggered tPA release was demonstrated, thus leading to effective thrombolysis under both static and flow conditions. Intravenous administration in vivo enabled prolonged circulation, enhanced thrombus accumulation and MR imaging. In a rat carotid thrombosis model, tPA@THPTN outperformed free tPA and non-targeted controls, achieving complete clot dissolution in 120 min with minimal hemorrhage risk. This biomimetic platform offers a potent theranostic strategy for thrombosis
A differentiable variational model for structural self-contact and fracture
No full textNumerical modelling of structural self-contact and crack propagation presents significant challenges due to the inherently discontinuous and non-differentiable
nature of the underlying physical phenomena. Traditional contact models demand explicit definition and tracking of contact points, while fracture models often rely on predefined crack initiation sites, sharp interfaces, and re-meshing. This study introduces a novel framework that overcomes these limitations within a unified and numerically stable variational formulation. The contact phenomenon is described through the hyperelastic third medium contact model and fracture is represented by a phase field. Structures are embedded in a third
medium, that stiffens under compression, enabling the transfer of forces between structural members. Crack propagation occurs in regions in which it is energetically favourable for the system to evolve toward a fully damaged state, specifically where the critical energy release rate is exceeded. Careful treatment is required when coupling the two phenomena, particularly concerning the void material behaviour. This work presents an efficient and differentiable numerical model that captures both nonlinear phenomena within a unified framework. This framework will allow designers and engineers to efficiently analyse complex nonlinear structural behaviours, previously requiring separate models that involved pre-defined crack initiation sites and contact points. Lastly, the differentiable nature of the model facilitates straightforward future integration into
topology optimisation pipelines, providing designers the ability to intentionally design for and leverage self-contact interactions and material failure as functional,
performance-enhancing features
Von Willebrand factor deficiency impairs angiogenesis via angiopoietin-2: relevance for gut angiodysplasia
No full textManagement of recurrent gastrointestinal (GI) bleeding is a clinical unmet need for patients with Von Willebrand disease (VWD) and is linked to the presence of gut vascular malformations (angiodysplasia). We previously demonstrated that von Willebrand factor (VWF) regulates angiogenesis and vascular integrity. VWF controls the storage of the angiogenesis regulator Angiopoietin-2 (Angpt-2) in endothelial cells (EC), suggesting a candidate for the genesis of angiodysplasia; however, no direct evidence of the role of Angpt-2 in VWF-dependent angiogenesis is available. Using VWF-deficient Human Umbilical Vein Endothelial Cells (HUVEC) and endothelial colony forming cells (ECFCs) from severe VWD patients, we find that loss of VWF results in increased Angpt-2 expression through the positive feedback loop Angpt-2-TIE2-AKT-FOXO1-Angpt-2. In the gut of VWF-deficient mice, Angpt-2 expression is increased whilst Angpt-1 expression is decreased, suggesting that VWF regulates the Angpt/Tie2 balance in the gut. Moreover, the intestinal vasculature in the jejunum of VWF-deficient mice appears abnormal, with hyper-sprouting and lumen formation defects. The findings reveal VWF-deficient mice as a model to study gut angiodysplasia. We investigate sprouting angiogenesis in vitro using a fibrin bead assay and find increased sprouting in VWF-deficient EC. We develop a 3D-microfluidic model of angiogenesis and find that ECFCs from severe VWD patients exhibit defective remodeling and abnormal lumen formation, reminiscent of the defects in the gut of VWF KO mice. Importantly, inhibition of Angpt-2 reduces sprouting in VWF-deficient HUVEC and normalises vascular remodeling in VWD ECFCs, suggesting that Angpt-2 inhibitors may be effective in VWD patients with GI bleeding and angiodysplasia.</jats:p
Chemical Engineering Research. Reports of the 4th year research projects in the Department of Chemical Engineering at Imperial College London. Volume 8
No full textThis volume of Chemical Engineering Research collects the unedited research project reports written by 4th-year undergraduates (Class of 2026) of the M.Eng. course on Chemical Engineering in the Department of Chemical Engineering at Imperial College London.
The research project spans one term (Autumn) during the last year of the career. It emphasises independence, the ability to plan and pursue original project work for an extended period, produce a high-quality report, and present the work to an audience using appropriate visual aids. Students are also expected to produce a literature review and situate their work in the context of prior art. The papers presented showcase the diversity and depth of some of the research streams in the department, but only touch on a small number of research groups and interests. For a complete description of the research at the department, the reader is referred to the departmental website¹.
The papers are presented in no particular order, each identified by a manuscript number. Some papers refer to appendices and/or supplementary materials that are too lengthy to include here. These files can be obtained directly from the supervisors (see the supervisor index at the end of the book). Some reports are missing and are under embargo as they contain confidential information.
The cover figure shows a microscopic (10X) image of alanine crystals in deionized water (from the work of Adam Adan and Aliff Rafiq, manuscript 13)
Dynamic multi-objective optimization algorithm based on historical collaborative strategy and interval prediction strategy
No full textThe file attached to this record is the author's final peer reviewed version. The Publisher's final version can be found by following the DOI link.Dynamic multi-objective optimization problems (DMOPs) involve scenarios where objective functions, decision variables, parameters, or other elements vary over time. An effective approach to address DMOPs is to integrate algorithms designed for static multi-objective optimization problems with dynamic response strategies. To improve the performance of these strategies in terms of both population diversity and convergence, this paper proposes a novel dynamic response strategy, the historical collaborative and interval prediction strategy (HCIPS). When confronted with environmental changes, we conduct a three-level population analysis: overall, historical, and individual. Firstly, the interval-based response strategy identifies interval partitioning of a population at time, enabling global localization of the predicted population and effectively preserving diverse population information. Secondly, the history-based response strategy guides the population movement by selecting optimal solutions from historical populations. Thirdly, the individual-based response strategy predicts individual positions by tracking the movement of key points. This serves as a crucial complement to the history-based response strategy, compensating for its primary drawback: a lack of sufficient historical data in the early stages of evolution. Experimental results indicate that the HCIPS offers advantages in solving DMOPs compared to past state-of-the-art algorithms