Publications from our Researchers

Several of our current PhD candidates and fellow researchers at the Data Science Institute have published, or in the proccess of publishing, papers to present their research.  

Search or filter publications

Filter by type:

Filter by publication type

Filter by year:



  • Showing results for:
  • Reset all filters

Search results

  • Conference paper
    Biffi C, Oktay O, Tarroni G, Bai W, De Marvao A, Doumou G, Rajchl M, Bedair R, Prasad S, Cook S, O’Regan D, Rueckert Det al., 2018,

    Learning interpretable anatomical features through deep generative models: Application to cardiac remodeling

    , International Conference On Medical Image Computing & Computer Assisted Intervention, Publisher: Springer, Pages: 464-471, ISSN: 0302-9743

    Alterations in the geometry and function of the heart define well-established causes of cardiovascular disease. However, current approaches to the diagnosis of cardiovascular diseases often rely on subjective human assessment as well as manual analysis of medical images. Both factors limit the sensitivity in quantifying complex structural and functional phenotypes. Deep learning approaches have recently achieved success for tasks such as classification or segmentation of medical images, but lack interpretability in the feature extraction and decision processes, limiting their value in clinical diagnosis. In this work, we propose a 3D convolutional generative model for automatic classification of images from patients with cardiac diseases associated with structural remodeling. The model leverages interpretable task-specific anatomic patterns learned from 3D segmentations. It further allows to visualise and quantify the learned pathology-specific remodeling patterns in the original input space of the images. This approach yields high accuracy in the categorization of healthy and hypertrophic cardiomyopathy subjects when tested on unseen MR images from our own multi-centre dataset (100%) as well on the ACDC MICCAI 2017 dataset (90%). We believe that the proposed deep learning approach is a promising step towards the development of interpretable classifiers for the medical imaging domain, which may help clinicians to improve diagnostic accuracy and enhance patient risk-stratification.

  • Conference paper
    Qin C, Bai W, Schlemper J, Petersen SE, Piechnik SK, Neubauer S, Rueckert Det al., 2018,

    Joint learning of motion estimation and segmentation for cardiac MR image sequences

    , International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), Publisher: Springer Verlag, Pages: 472-480

    Cardiac motion estimation and segmentation play important roles in quantitatively assessing cardiac function and diagnosing cardiovascular diseases. In this paper, we propose a novel deep learning method for joint estimation of motion and segmentation from cardiac MR image sequences. The proposed network consists of two branches: a cardiac motion estimation branch which is built on a novel unsupervised Siamese style recurrent spatial transformer network, and a cardiac segmentation branch that is based on a fully convolutional network. In particular, a joint multi-scale feature encoder is learned by optimizing the segmentation branch and the motion estimation branch simultaneously. This enables the weakly-supervised segmentation by taking advantage of features that are unsupervisedly learned in the motion estimation branch from a large amount of unannotated data. Experimental results using cardiac MlRI images from 220 subjects show that the joint learning of both tasks is complementary and the proposed models outperform the competing methods significantly in terms of accuracy and speed.

  • Conference paper
    Schlemper J, Oktay O, Bai W, Castro DC, Duan J, Qin C, Hajnal JV, Rueckert Det al., 2018,

    Cardiac MR segmentation from undersampled k-space using deep latent representation learning

    , International Conference On Medical Image Computing & Computer Assisted Intervention, Publisher: Springer, Cham, Pages: 259-267, ISSN: 0302-9743

    © Springer Nature Switzerland AG 2018. Reconstructing magnetic resonance imaging (MRI) from undersampled k-space enables the accelerated acquisition of MRI but is a challenging problem. However, in many diagnostic scenarios, perfect reconstructions are not necessary as long as the images allow clinical practitioners to extract clinically relevant parameters. In this work, we present a novel deep learning framework for reconstructing such clinical parameters directly from undersampled data, expanding on the idea of application-driven MRI. We propose two deep architectures, an end-to-end synthesis network and a latent feature interpolation network, to predict cardiac segmentation maps from extremely undersampled dynamic MRI data, bypassing the usual image reconstruction stage altogether. We perform a large-scale simulation study using UK Biobank data containing nearly 1000 test subjects and show that with the proposed approaches, an accurate estimate of clinical parameters such as ejection fraction can be obtained from fewer than 10 k-space lines per time-frame.

  • Conference paper
    Bai W, Suzuki H, Qin C, Tarroni G, Oktay O, Matthews PM, Rueckert Det al., 2018,

    Recurrent neural networks for aortic image sequence segmentation with sparse annotations

    , International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), ISSN: 0302-9743

    Segmentation of image sequences is an important task in medical image analysis, which enables clinicians to assess the anatomy and function of moving organs. However, direct application of a segmentation algorithm to each time frame of a sequence may ignore the temporal continuity inherent in the sequence. In this work, we propose an image sequence segmentation algorithm by combining a fully convolutional network with a recurrent neural network, which incorporates both spatial and temporal information into the segmentation task. A key challenge in training this network is that the available manual annotations are temporally sparse, which forbids end-to-end training. We address this challenge by performing non-rigid label propagation on the annotations and introducing an exponentially weighted loss function for training. Experiments on aortic MR image sequences demonstrate that the proposed method significantly improves both accuracy and temporal smoothness of segmentation, compared to a baseline method that utilises spatial information only. It achieves an average Dice metric of 0.960 for the ascending aorta and 0.953 for the descending aorta.

  • Journal article
    Bai W, Sinclair M, Tarroni G, Oktay O, Rajchl M, Vaillant G, Lee AM, Aung N, Lukaschuk E, Sanghvi MM, Zemrak F, Fung K, Paiva JM, Carapella V, Kim YJ, Suzuki H, Kainz B, Matthews PM, Petersen SE, Piechnik SK, Neubauer S, Glocker B, Rueckert Det al., 2018,

    Automated cardiovascular magnetic resonance image analysis with fully convolutional networks

    Cardiovascular magnetic resonance (CMR) imaging is a standard imagingmodality for assessing cardiovascular diseases (CVDs), the leading cause ofdeath globally. CMR enables accurate quantification of the cardiac chambervolume, ejection fraction and myocardial mass, providing information fordiagnosis and monitoring of CVDs. However, for years, clinicians have beenrelying on manual approaches for CMR image analysis, which is time consumingand prone to subjective errors. It is a major clinical challenge toautomatically derive quantitative and clinically relevant information from CMRimages. Deep neural networks have shown a great potential in image patternrecognition and segmentation for a variety of tasks. Here we demonstrate anautomated analysis method for CMR images, which is based on a fullyconvolutional network (FCN). The network is trained and evaluated on alarge-scale dataset from the UK Biobank, consisting of 4,875 subjects with93,500 pixelwise annotated images. The performance of the method has beenevaluated using a number of technical metrics, including the Dice metric, meancontour distance and Hausdorff distance, as well as clinically relevantmeasures, including left ventricle (LV) end-diastolic volume (LVEDV) andend-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolicvolume (RVEDV) and end-systolic volume (RVESV). By combining FCN with alarge-scale annotated dataset, the proposed automated method achieves a highperformance on par with human experts in segmenting the LV and RV on short-axisCMR images and the left atrium (LA) and right atrium (RA) on long-axis CMRimages.

  • Conference paper
    Robinson R, Oktay O, Bai W, Valindria V, Sanghvi MM, Aung N, Paiva JM, Zemrak F, Fung K, Lukaschuk E, Lee AM, Carapella V, Kim YJ, Kainz B, Piechnik SK, Neubauer S, Petersen SE, Page C, Rueckert D, Glocker Bet al., 2018,

    Real-time prediction of segmentation quality

    , International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Publisher: Springer Verlag, Pages: 578-585, ISSN: 0302-9743

    Recent advances in deep learning based image segmentationmethods have enabled real-time performance with human-level accuracy.However, occasionally even the best method fails due to low image qual-ity, artifacts or unexpected behaviour of black box algorithms. Beingable to predict segmentation quality in the absence of ground truth is ofparamount importance in clinical practice, but also in large-scale studiesto avoid the inclusion of invalid data in subsequent analysis.In this work, we propose two approaches of real-time automated qualitycontrol for cardiovascular MR segmentations using deep learning. First,we train a neural network on 12,880 samples to predict Dice SimilarityCoefficients (DSC) on a per-case basis. We report a mean average error(MAE) of 0.03 on 1,610 test samples and 97% binary classification accu-racy for separating low and high quality segmentations. Secondly, in thescenario where no manually annotated data is available, we train a net-work to predict DSC scores from estimated quality obtained via a reversetesting strategy. We report an MAE = 0.14 and 91% binary classifica-tion accuracy for this case. Predictions are obtained in real-time which,when combined with real-time segmentation methods, enables instantfeedback on whether an acquired scan is analysable while the patient isstill in the scanner. This further enables new applications of optimisingimage acquisition towards best possible analysis results.

  • Conference paper
    Duan J, Schlemper J, Bai W, Dawes TJW, Bello G, Doumou G, De Marvao A, O'Regan DP, Rueckert Det al., 2018,

    Deep Nested Level Sets: Fully Automated Segmentation of Cardiac MR Images in Patients with Pulmonary Hypertension

    , International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), Pages: 595-603, ISSN: 0302-9743
  • Journal article
    Gómez-Romero J, Molina-Solana M, Ros M, Ruiz MD, Martin-Bautista MJet al., 2018,

    Comfort as a service: a new paradigm for residential environmental quality control

    , Sustainability, Vol: 10, ISSN: 1937-0709

    This paper introduces the concept of Comfort as a Service (CaaS), a new energy supply paradigm for providing comfort to residential customers. CaaS takes into account the available passive and active elements, the external factors that affect energy consumption and associated costs, and occupants' behaviors to generate optimal control strategies for the domestic equipment automatically. As a consequence, it releases building occupants from operating the equipment, which gives rise to a disruption of the traditional model of paying per consumed energy in favor of a model of paying per provided comfort. In the paper, we envision a realization of CaaS based on several technologies such as ambient intelligence, big data, cloud computing and predictive computing. We discuss the opportunities and the barriers of CaaS-centered business and exemplify the potential of CaaS deployments by quantifying the expected energy savings achieved after limiting occupants' control over the air conditioning system in a test scenario.

  • Journal article
    Song J, Fan S, Lin W, Mottet L, Woodward H, Wykes MD, Arcucci R, Xiao D, Debay J-E, ApSimon H, Aristodemou E, Birch D, Carpentieri M, Fang F, Herzog M, Hunt GR, Jones RL, Pain C, Pavlidis D, Robins AG, Short CA, Linden PFet al., 2018,

    Natural ventilation in cities: the implications of fluid mechanics

    , BUILDING RESEARCH AND INFORMATION, Vol: 46, Pages: 809-828, ISSN: 0961-3218
  • Journal article
    Takahashi K, Pavlidis S, Ng Kee Kwong F, Hoda U, Rossios C, Sun K, Loza M, Baribaud F, Chanez P, Fowler SJ, Horvath I, Montuschi P, Singer F, Musial J, Dahlen B, Dahlen SE, Krug N, Sandstrom T, Shaw DE, Lutter R, Bakke P, Fleming LJ, Howarth PH, Caruso M, Sousa AR, Corfield J, Auffray C, De Meulder B, Lefaudeux D, Djukanovic R, Sterk PJ, Guo Y, Adcock I, Chung KFet al., 2018,

    Sputum proteomics and airway cell transcripts of current and ex-smokers with severe asthma in U-BIOPRED: an exploratory analysis

    , European Respiratory Journal, Vol: 51, ISSN: 0903-1936

    Background: Severe asthma patients with a significant smoking history have airflow obstruction with reported neutrophilia. We hypothesise that multi1omic analysis will enable the definition of smoking and ex1smoking severe asthma molecular phenotypes.Methods The U1BIOPRED severe asthma patients containing current1smokers (CSA), ex1smokers (ESA), non1smokers (NSA) and healthy non1smokers (NH) was examined. Blood and sputum cell counts, fractional exhaled nitric oxide and spirometry were obtained. Exploratory proteomic analysis of sputum supernatants and transcriptomic analysis of bronchial brushings, biopsies and sputum cells was performed. Results Colony stimulating factor (CSF)2 protein levels were increased in CSA sputum supernatants with azurocidin 1, neutrophil elastase and CXCL8 upregulated in ESA. Phagocytosis and innate immune pathways were associated with neutrophilic inflammation in ESA. Gene Set Variation Analysis of bronchial epithelial cell transcriptome from CSA showed enrichment of xenobiotic metabolism, oxidative stress and endoplasmic reticulum stress compared to other groups. CXCL5 and matrix metallopeptidase 12 genes were upregulated in ESA and the epithelial protective genes, mucin 2 and cystatin SN, were downregulated. Conclusion Despite little difference in clinical characteristics, CSA were distinguishable from ESA subjects at the sputum proteomic level with CSA having increased CSF2 expression and ESA patients showed sustained loss of epithelial barrier processes.

  • Journal article
    Hekking PP, Loza MJ, Pavlidis S, De Meulder B, Lefaudeux D, Baribaud F, Auffray C, Wagener AH, Brinkman P, Lutter R, Bansal AT, Sousa AR, Bates S, Pandis Y, Fleming LJ, Shaw DE, Fowler SJ, Guo Y, Meiser A, Sun K, Corfield J, Howarth P, Bel EH, Adcock IM, Chung KF, Djukanovic R, Sterk PJ, U-BIOPRED Study Groupet al., 2017,

    Transcriptomic gene signatures associated with persistent airflow limitation in patients with severe asthma

    , European Respiratory Journal, Vol: 50, ISSN: 1399-3003

    Rationale:A proportion of severe asthma patients suffers from persistent airflow limitation, often associated with more symptoms and exacerbations. Little is known about the underlying mechanisms. Aiming for discovery of unexplored potential mechanisms, we used Gene Set Variation Analysis (GSVA), a sensitive technique that can detect underlying pathways in heterogeneous samples. Methods: Severe asthma patients from the U-BIOPRED cohort with persistent airflow limitation (post-bronchodilator FEV1/FVC ratio < lower limit of normal) were compared to those without persistent airflow limitation. Gene expression was assessed on the total RNA of sputum cells, nasal brushings and endobronchial brushings and biopsies. GSVA was applied to identify differentially-enriched pre-defined gene signatures based on all available gene expression publications and data on airways disease.Results: Differentially-enriched gene signatures were identified in nasal brushings (1), sputum (9), bronchial brushings (1) and bronchial biopsies (4), that were associated with response to inhaled steroids, eosinophils, IL-13, IFN-alpha, specific CD4+ T-cells and airway remodeling.Conclusion: Persistent airflow limitation in severe asthma has distinguishable underlying gene networks that are associated with treatment, inflammatory pathways and airway remodeling. These results point towards targets for the therapy of persistent airflow limitation in severe asthma.

  • Journal article
    Hekking PP, Loza MJ, Pavlidis S, De Meulder B, Lefaudeux D, Baribaud F, Auffray C, Wagener A, Brinkman P, Lutter I, Bansal A, Sousa A, Bates S, Pandis Y, Fleming L, Shaw DE, Fowler SJ, Guo Y, Meiser A, Sun K, Corfield J, Howarth P, Bel EH, Adcock IM, Chung KF, Djukanovic R, Sterk PJ, U-BIOPRED Study Groupet al., 2017,

    Pathway discovery using transcriptomic profiles in adult-onset severe asthma

    , Journal of Allergy and Clinical Immunology, Vol: 141, Pages: 1280-1290, ISSN: 1097-6825

    RationaleAdult-onset severe asthma is characterized by highly symptomatic disease despite high intensity asthma treatments. Understanding of the underlying pathways of this heterogeneous disease needed for the development of targeted treatments. Gene Set Variation Analysis (GSVA) is a statistical technique to identify gene profiles in heterogeneous samples.ObjectiveTo identify gene profiles associated with adult-onset severe asthma.MethodsThis was a cross-sectional, observational study in which adult patients with adult-onset of asthma (defined as starting at ≥18yrs old) as compared to childhood-onset severe asthma (<18 yrs) were selected from the U-BIOPRED cohort. Gene expression was assessed on the total RNA of induced sputum (n=83), nasal brushings (n=41), and endobronchial brushings (n=65) and biopsies (n=47) (Affymetrix HT HG-U133+ PM). GSVA was used to identify differentially enriched pre-defined gene signatures of leukocyte lineage, inflammatory and induced lung injury pathways.ResultsSignificant differentially enriched gene signatures in patients with adult-onset as compared to childhood-onset severe asthma were identified in nasal brushings (5 signatures), sputum (3 signatures) and endobronchial brushings (6 signatures). Signatures associated with eosinophilic airway inflammation, mast cells and group 3 innate lymphoid cells (ILC3) were more enriched in adult-onset severe asthma, whereas signatures associated with induced lung injury were less enriched in adult-onset severe asthma.ConclusionsAdult-onset severe asthma is characterized by inflammatory pathways involving eosinophils, mast cells and ILC3s. These pathways could represent useful targets for the treatment of adult-onset severe asthma.

  • Journal article
    Rossios C, Pavlidis S, Hoda U, Kuo CH, Wiegman C, Russell K, Sun K, Loza MJ, Baribaud F, Durham AL, Ojo O, Lutter R, Rowe A, Bansal A, Auffray C, Sousa A, Corfield J, Djukanovic R, Guo Y, Sterk PJ, Chung KF, Adcock IM, Unbiased Biomarkers for the Prediction of Respiratory Diseases Outcomes U-BIOPRED Consortia Project Teamet al., 2017,

    Sputum transcriptomics reveal upregulation of IL-1 receptor family members in patients with severe asthma

    , Journal of Allergy and Clinical Immunology, Vol: 141, Pages: 560-570, ISSN: 1097-6825

    BACKGROUND: Sputum analysis in asthmatic patients is used to define airway inflammatory processes and might guide therapy. OBJECTIVE: We sought to determine differential gene and protein expression in sputum samples from patients with severe asthma (SA) compared with nonsmoking patients with mild/moderate asthma. METHODS: Induced sputum was obtained from nonsmoking patients with SA, smokers/ex-smokers with severe asthma, nonsmoking patients with mild/moderate asthma (MMAs), and healthy nonsmoking control subjects. Differential cell counts, microarray analysis of cell pellets, and SOMAscan analysis of sputum analytes were performed. CRID3 was used to inhibit the inflammasome in a mouse model of SA. RESULTS: Eosinophilic and mixed neutrophilic/eosinophilic inflammation were more prevalent in patients with SA compared with MMAs. Forty-two genes probes were upregulated (>2-fold) in nonsmoking patients with severe asthma compared with MMAs, including IL-1 receptor (IL-1R) family and nucleotide-binding oligomerization domain, leucine-rich repeat and pyrin domain containing 3 (NRLP3) inflammasome members (false discovery rate < 0.05). The inflammasome proteins nucleotide-binding oligomerization domain, leucine rich repeat and pyrin domain containing 1 (NLRP1), NLRP3, and nucleotide-binding oligomerization domain (NOD)-like receptor C4 (NLRC4) were associated with neutrophilic asthma and with sputum IL-1β protein levels, whereas eosinophilic asthma was associated with an IL-13-induced TH2 signature and IL-1 receptor-like 1 (IL1RL1) mRNA expression. These differences were sputum specific because no activation of NLRP3 or enrichment of IL-1R family genes in bronchial brushings or biopsy specimens in patients with SA was observed. Expression of NLRP3 and of the IL-1R family genes was validated in the Airway Disease Endotyping for Personalized Therapeutics cohort. Inflammasome inhibition using CRID3 prevented airway hyperresponsiveness and airway inflammati

  • Journal article
    Jahani E, Sundsøy P, Bjelland J, Bengtsson L, Pentland AS, de Montjoye Y-Aet al., 2017,

    Improving official statistics in emerging markets using machine learning and mobile phone data

    , EPJ Data Science, Vol: 6, ISSN: 2193-1127

    Mobile phones are one of the fastest growing technologies in the developing world with global penetration rates reaching 90%. Mobile phone data, also called CDR, are generated everytime phones are used and recorded by carriers at scale. CDR have generated groundbreaking insights in public health, official statistics, and logistics. However, the fact that most phones in developing countries are prepaid means that the data lacks key information about the user, including gender and other demographic variables. This precludes numerous uses of this data in social science and development economic research. It furthermore severely prevents the development of humanitarian applications such as the use of mobile phone data to target aid towards the most vulnerable groups during crisis. We developed a framework to extract more than 1400 features from standard mobile phone data and used them to predict useful individual characteristics and group estimates. We here present a systematic cross-country study of the applicability of machine learning for dataset augmentation at low cost. We validate our framework by showing how it can be used to reliably predict gender and other information for more than half a million people in two countries. We show how standard machine learning algorithms trained on only 10,000 users are sufficient to predict individual’s gender with an accuracy ranging from 74.3 to 88.4% in a developed country and from 74.5 to 79.7% in a developing country using only metadata. This is significantly higher than previous approaches and, once calibrated, gives highly accurate estimates of gender balance in groups. Performance suffers only marginally if we reduce the training size to 5,000, but significantly decreases in a smaller training set. We finally show that our indicators capture a large range of behavioral traits using factor analysis and that the framework can be used to predict other indicators of vulnerability such as age or socio-economic status. M

  • Journal article
    Steele JE, Sundsoy PR, Pezzulo C, Alegana VA, Bird TJ, Blumenstock J, Bjelland J, Engo-Monsen K, de Montjoye YKJV, Iqbal AM, Hadiuzzaman KN, Lu X, Wetter E, Tatem AJ, Bengtsson Let al., 2017,

    Mapping poverty using mobile phone and satellite data

    , Journal of the Royal Society Interface, Vol: 14, ISSN: 1742-5689

    Poverty is one of the most important determinants of adverse health outcomesglobally, a major cause of societal instability and one of the largest causes of losthuman potential. Traditional approaches to measuring and targeting povertyrely heavily on census data, which in most low- and middle-income countries(LMICs) are unavailable or out-of-date.Alternate measures are needed to comp-lement and update estimates between censuses. This study demonstrates howpublic and private data sources that are commonly available for LMICs can beused to provide novel insight into the spatial distribution of poverty. We evalu-ate the relative value of modelling three traditional poverty measures usingaggregate data from mobile operators and widely available geospatial data.Taken together, models combining these data sources providethebest predictivepower (highestr2¼0.78) and lowest error, but generally models employingmobile data only yield comparable results, offering the potential to measurepoverty more frequently and at finer granularity. Stratifying models intourban and rural areas highlights the advantage of using mobile data in urbanareas and different data in different contexts. The findings indicate the possibilityto estimate and continually monitor poverty rates at high spatial resolution incountries with limited capacity to support traditional methods of datacollection.

This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.

Request URL: Request URI: /respub/WEB-INF/jsp/search-t4-html.jsp Query String: id=607&limit=15&page=6&respub-action=search.html Current Millis: 1621326561686 Current Time: Tue May 18 09:29:21 BST 2021