Catalogue Search | MBRL
Are you sure you want to remove the book from the shelf?
{{itemTitle}}
2,632
result(s) for
"Domínguez, Manuel"
Sort by:
Deep Learning System for COVID-19 Diagnosis Aid Using X-ray Pulmonary Images
The spread of the SARS-CoV-2 virus has made the COVID-19 disease a worldwide epidemic. The most common tests to identify COVID-19 are invasive, time consuming and limited in resources. Imaging is a non-invasive technique to identify if individuals have symptoms of disease in their lungs. However, the diagnosis by this method needs to be made by a specialist doctor, which limits the mass diagnosis of the population. Image processing tools to support diagnosis reduce the load by ruling out negative cases. Advanced artificial intelligence techniques such as Deep Learning have shown high effectiveness in identifying patterns such as those that can be found in diseased tissue. This study analyzes the effectiveness of a VGG16-based Deep Learning model for the identification of pneumonia and COVID-19 using torso radiographs. Results show a high sensitivity in the identification of COVID-19, around 100%, and with a high degree of specificity, which indicates that it can be used as a screening test. AUCs on ROC curves are greater than 0.9 for all classes considered.
Journal Article
Review on V2X, I2X, and P2X Communications and Their Applications: A Comprehensive Analysis over Time
Smart cities are ecosystems where novel ideas and emerging technologies meet to improve economy, environment, governance, living, and mobility. One of the pillars of smart cities is transport, with the improvement of mobility and the reduction of traffic accidents being some of the current key challenges. With this purpose, this manuscript reviews the state-of-the-art of communications and applications in which different actors of the road are involved. Thus, the objectives of this survey are intended to determine who, when, and about what is being researched around smart cities. Particularly, the goal is to situate the focus of scientific and industrial progress on V2X, I2X, and P2X communication to establish a taxonomy that reduces ambiguous acronyms around the communication between vehicles, infrastructure, and pedestrians, as well as to determine what the trends and future technologies are that will lead to more powerful applications. To this end, this literature review article presents a comprehensive study including a representative collection of the 100 most cited papers and patents published in the literature together with a statistical bibliometric analysis of 14,364 keywords over 3422 contributions between 1997 and 2018. As a result, this work provides a technological profile considering different dimensions along the paper, such as the type of communication, use case, country, organization, terminology, and year.
Journal Article
Sustainability and environmental impact of fused deposition modelling (FDM) technologies
This paper consists in a review of environmental issues of additive manufacturing technologies, mainly in those related to fused deposition modelling. The versatility, reduction of equipment costs due to patents expiring and the great flexibility offered by 3D printers have driven the amazing increase of these technologies in the last years. On the other hand, the democratization of additive manufacturing also poses some issues regarding environment; it is important to also have into account the potential effects of these technologies in the environment, coming from energy consumption, materials, and wastes produced. A review of different research works dealing with environmental impact of additive manufacturing, such as products’ life cycle assessment, energy and materials consumption, and particles and gases releases (mainly due to health issues), has been performed. The assessment performed has helped highlighting the importance of environmental issues in additive manufacturing, according to the number of published papers. The main findings are the importance of establishing a method for applying eco-design principles taking into account the specific features of additive manufacturing.
Journal Article
Distinguishing butchery cut marks from crocodile bite marks through machine learning methods
All models of evolution of human behaviour depend on the correct identification and interpretation of bone surface modifications (BSM) on archaeofaunal assemblages. Crucial evolutionary features, such as the origin of stone tool use, meat-eating, food-sharing, cooperation and sociality can only be addressed through confident identification and interpretation of BSM, and more specifically, cut marks. Recently, it has been argued that linear marks with the same properties as cut marks can be created by crocodiles, thereby questioning whether secure cut mark identifications can be made in the Early Pleistocene fossil record. Powerful classification methods based on multivariate statistics and machine learning (ML) algorithms have previously successfully discriminated cut marks from most other potentially confounding BSM. However, crocodile-made marks were marginal to or played no role in these comparative analyses. Here, for the first time, we apply state-of-the-art ML methods on crocodile linear BSM and experimental butchery cut marks, showing that the combination of multivariate taphonomy and ML methods provides accurate identification of BSM, including cut and crocodile bite marks. This enables empirically-supported hominin behavioural modelling, provided that these methods are applied to fossil assemblages.
Journal Article
CLoDSA: a tool for augmentation in classification, localization, detection, semantic segmentation and instance segmentation tasks
Background
Deep learning techniques have been successfully applied to bioimaging problems; however, these methods are highly data demanding. An approach to deal with the lack of data and avoid overfitting is the application of
data augmentation
, a technique that generates new training samples from the original dataset by applying different kinds of transformations. Several tools exist to apply data augmentation in the context of image classification, but it does not exist a similar tool for the problems of localization, detection, semantic segmentation or instance segmentation that works not only with 2 dimensional images but also with multi-dimensional images (such as stacks or videos).
Results
In this paper, we present a generic strategy that can be applied to automatically augment a dataset of images, or multi-dimensional images, devoted to classification, localization, detection, semantic segmentation or instance segmentation. The augmentation method presented in this paper has been implemented in the open-source package CLoDSA. To prove the benefits of using CLoDSA, we have employed this library to improve the accuracy of models for Malaria parasite classification, stomata detection, and automatic segmentation of neural structures.
Conclusions
CLoDSA is the first, at least up to the best of our knowledge, image augmentation library for object classification, localization, detection, semantic segmentation, and instance segmentation that works not only with 2 dimensional images but also with multi-dimensional images.
Journal Article
Augmented and Virtual Reality Evolution and Future Tendency
Augmented reality and virtual reality technologies are increasing in popularity. Augmented reality has thrived to date mainly on mobile applications, with games like Pokémon Go or the new Google Maps utility as some of its ambassadors. On the other hand, virtual reality has been popularized mainly thanks to the videogame industry and cheaper devices. However, what was initially a failure in the industrial field is resurfacing in recent years thanks to the technological improvements in devices and processing hardware. In this work, an in-depth study of the different fields in which augmented and virtual reality have been used has been carried out. This study focuses on conducting a thorough scoping review focused on these new technologies, where the evolution of each of them during the last years in the most important categories and in the countries most involved in these technologies will be analyzed. Finally, we will analyze the future trend of these technologies and the areas in which it is necessary to investigate to further integrate these technologies into society.
Journal Article
Smouldering multiple sclerosis: the ‘real MS’
Using a philosophical approach or deductive reasoning, we challenge the dominant clinico-radiological worldview that defines multiple sclerosis (MS) as a focal inflammatory disease of the central nervous system (CNS). We provide a range of evidence to argue that the ‘real MS’ is in fact driven primarily by a smouldering pathological disease process. In natural history studies and clinical trials, relapses and focal activity revealed by magnetic resonance imaging (MRI) in MS patients on placebo or on disease-modifying therapies (DMTs) were found to be poor predictors of long-term disease evolution and were dissociated from disability outcomes. In addition, the progressive accumulation of disability in MS can occur independently of relapse activity from early in the disease course. This scenario is underpinned by a more diffuse smouldering pathological process that may affect the entire CNS. Many putative pathological drivers of smouldering MS can be potentially modified by specific therapeutic strategies, an approach that may have major implications for the management of MS patients. We hypothesise that therapeutically targeting a state of ‘no evident inflammatory disease activity’ (NEIDA) cannot sufficiently prevent disability accumulation in MS, meaning that treatment should also focus on other brain and spinal cord pathological processes contributing to the slow loss of neurological function. This should also be complemented with a holistic approach to the management of other systemic disease processes that have been shown to worsen MS outcomes.
Journal Article
Wearable Fall Detector Using Recurrent Neural Networks
Falls have become a relevant public health issue due to their high prevalence and negative effects in elderly people. Wearable fall detector devices allow the implementation of continuous and ubiquitous monitoring systems. The effectiveness for analyzing temporal signals with low energy consumption is one of the most relevant characteristics of these devices. Recurrent neural networks (RNNs) have demonstrated a great accuracy in some problems that require analyzing sequential inputs. However, getting appropriate response times in low power microcontrollers remains a difficult task due to their limited hardware resources. This work shows a feasibility study about using RNN-based deep learning models to detect both falls and falls’ risks in real time using accelerometer signals. The effectiveness of four different architectures was analyzed using the SisFall dataset at different frequencies. The resulting models were integrated into two different embedded systems to analyze the execution times and changes in the model effectiveness. Finally, a study of power consumption was carried out. A sensitivity of 88.2% and a specificity of 96.4% was obtained. The simplest models reached inference times lower than 34 ms, which implies the capability to detect fall events in real-time with high energy efficiency. This suggests that RNN models provide an effective method that can be implemented in low power microcontrollers for the creation of autonomous wearable fall detection systems in real-time.
Journal Article
Complex regulation in a Comamonas platform for diverse aromatic carbon metabolism
Critical to a sustainable energy future are microbial platforms that can process aromatic carbons from the largely untapped reservoir of lignin and plastic feedstocks.
Comamonas
species present promising bacterial candidates for such platforms because they can use a range of natural and xenobiotic aromatic compounds and often possess innate genetic constraints that avoid competition with sugars. However, the metabolic reactions of these species are underexplored, and the regulatory mechanisms are unknown. Here we identify multilevel regulation in the conversion of lignin-related natural aromatic compounds, 4-hydroxybenzoate and vanillate, and the plastics-related xenobiotic aromatic compound, terephthalate, in
Comamonas testosteroni
KF-1. Transcription-level regulation controls initial catabolism and cleavage, but metabolite-level thermodynamic regulation governs fluxes in central carbon metabolism. Quantitative
13
C mapping of tricarboxylic acid cycle and cataplerotic reactions elucidates key carbon routing not evident from enzyme abundance changes. This scheme of transcriptional activation coupled with metabolic fine-tuning challenges outcome predictions during metabolic manipulations.
Aromatic carbon fluxes for the metabolism of lignin and plastics derivatives in
Comamonas testosteroni
KF-1 are controlled by pathway-specific regulation at transcript, protein or metabolite levels. The combination of transcriptional activation and metabolic fine-tuning complicates predictions of modulated carbon and energy fluxes during metabolic engineering strategies.
Journal Article