Breast Delineation in Full-Field Digital Mammography Using the Segment Anything Model.

Breast cancer is a major health concern worldwide. Mammography, a cost-effective and accurate tool, is crucial in combating this issue. However, low contrast, noise, and artifacts can limit the diagnostic capabilities of radiologists. Computer-Aided Diagnosis (CAD) systems have been developed to overcome these challenges, with the accurate outlining of the breast being a critical step for further analysis. This study introduces the SAM-breast model, an adaptation of the Segment Anything Model (SAM) for segmenting the breast region in mammograms. This method enhances the delineation of the breast and the exclusion of the pectoral muscle in both medio lateral-oblique (MLO) and cranio-caudal (CC) views. We trained the models using a large, multi-center proprietary dataset of 2492 mammograms. The proposed SAM-breast model achieved the highest overall Dice Similarity Coefficient (DSC) of 99.22% ± 1.13 and Intersection over Union (IoU) 98.48% ± 2.10 over independent test images from five different datasets (two proprietary and three publicly available). The results are consistent across the different datasets, regardless of the vendor or image resolution. Compared with other baseline and deep learning-based methods, the proposed method exhibits enhanced performance. The SAM-breast model demonstrates the power of the SAM to adapt when it is tailored to specific tasks, in this case, the delineation of the breast in mammograms. Comprehensive evaluations across diverse datasets-both private and public-attest to the method's robustness, flexibility, and generalization capabilities.

Prediction of 30-day unplanned hospital readmission through survival analysis.

Background and Objective: Unplanned hospital readmissions are a severe and recurrent problem that affects all health systems. Estimating the risk of being readmitted the following days after discharge is difficult since many heterogeneous factors can influence this. The extensive work concerning this problem proposes solutions mostly based on classification machine-learning models. Survival analysis methods could make a better match with the assessment of readmission risk and are yet to become well-established in this field. Methods: We compare different statistical and machine learning survival analysis models trained with right-censored all-cause hospital admission data with covariates available at the moment of discharge. The main focus is on tree-ensemble regression methods based on the assumption of proportional hazards. These models are more thoroughly evaluated at a 30-day time period after discharge, although the actual prediction could be set to any time up to 90 days. Results: The mean performance obtained by each of the proposed survival models ranges from 0.707 to 0.716 C-Index and 0.709 to 0.72 ROC-AUC at a 30-day time period after discharge. The model with the lower performance on both metrics was Cox Proportional Hazards, while the model marking the upper end on both ranges is an XGBoost Regression model with a Cox objective function. Conclusions: Our findings indicate that survival models perform well addressing the hospital readmission problem, machine-learning models getting the edge over statistical methods. There seems to be an improvement over classification models when attempting to predict at a 30-day period since discharge, perhaps due to a better handling of cases nearing the 30-day boundary. Some preprocessing steps, such as limiting the observation period to 90 days after discharge, are also highlighted since they resulted in a performance boost.

Analysis of the Localization of Schizosaccharomyces pombe Glucan Synthases in the Presence of the Antifungal Agent Caspofungin.

In recent years, invasive fungal infections have emerged as a common source of infections in immunosuppressed patients. All fungal cells are surrounded by a cell wall that is essential for cell integrity and survival. It prevents cell death and lysis resulting from high internal turgor pressure. Since the cell wall is not present in animal cells, it is an ideal target for selective invasive fungal infection treatments. The antifungal family known as echinocandins, which specifically inhibit the synthesis of the cell wall ß(13)glucan, has been established as an alternative treatment for mycoses. To explore the mechanism of action of these antifungals, we analyzed the cell morphology and glucan synthases localization in Schizosaccharomyces pombe cells during the initial times of growth in the presence of the echinocandin drug caspofungin. S. pombe are rod-shaped cells that grow at the poles and divide by a central division septum. The cell wall and septum are formed by different glucans, which are synthesized by four essential glucan synthases: Bgs1, Bgs3, Bgs4, and Ags1. Thus, S. pombe is not only a perfect model for studying the synthesis of the fungal ß(1-3)glucan, but also it is ideal for examining the mechanisms of action and resistance of cell wall antifungals. Herein, we examined the cells in a drug susceptibility test in the presence of either lethal or sublethal concentrations of caspofungin, finding that exposure to the drug for long periods at high concentrations (>10 µg/mL) induced cell growth arrest and the formation of rounded, swollen, and dead cells, whereas low concentrations (<10 µg/mL) permitted cell growth with a mild effect on cell morphology. Interestingly, short-term treatments with either high or low concentrations of the drug induced effects contrary to those observed in the susceptibility tests. Thus, low drug concentrations induced a cell death phenotype that was not observed at high drug concentrations, which caused transient fungistatic cell growth arrest. After 3 h, high concentrations of the drug caused the following: (i) a decrease in the GFP-Bgs1 fluorescence level; (ii) altered locations of Bgs3, Bgs4, and Ags1; and (iii) a simultaneous accumulation of cells with calcofluor-stained incomplete septa, which at longer times resulted in septation uncoupling from plasma membrane ingression. The incomplete septa revealed with calcofluor were found to be complete when observed via the membrane-associated GFP-Bgs or Ags1-GFP. Finally, we found that the accumulation of incomplete septa depended on Pmk1, the last kinase of the cell wall integrity pathway.

SEQENS: An ensemble method for relevant gene identification in microarray data.

This paper describes an ensemble feature identification algorithm called SEQENS, and measures its capability to identify the relevant variables in a case-control study using a genetic expression microarray dataset. SEQENS uses Sequential Feature Search on multiple sample splitting to select variables showing stronger relation with the target, and a variable relevance ranking is finally produced. Although designed for feature identification, SEQENS could also serve as a basis for feature selection (classifier optimisation). Cliff, a ranking evaluation metric is also presented and used to assess the feature identification algorithms when a groundtruth of relevant variables is available. To test performance, three types of synthetic groundtruths emulating fictitious diseases are generated from ten randomly chosen variables following different target pattern distributions using the E-MTAB-3732 dataset. Several sample-to-dimensionality ratios ranging from 300 to 3,000 observations and 854 to 54,675 variables are explored. SEQENS is compared with other feature selection or identification state-of-the-art methods. On average, the proposed algorithm identifies better the relevant genes and exhibits a stronger stability. The algorithm is available to the community.

Evaluation of Zearalenones and Their Metabolites in Chicken, Pig and Lamb Liver Samples.

Zearalenone (ZON), zearalanone (ZAN) and their phase I metabolites: α-zearalenol (α-ZOL), ß-zearalenol (ß-ZOL), α-zearalalanol (α-ZAL) and ß-zearalalanol (ß-ZAL) are compounds with estrogenic activity that are metabolized and distributed by the circulatory system in animals and can access the food chain through meat products from livestock. Furthermore, biomonitoring of zearalenones in biological matrices can provide useful information to directly assess mycotoxin exposure; therefore, their metabolites may be suitable biomarkers. The aim of this study was to determine the presence of ZON, ZAN and their metabolites in alternative biological matrices, such as liver, from three different animals: chicken, pig and lamb, in order to evaluate their exposure. A solid-liquid extraction procedure coupled to a GC-MS/MS analysis was performed. The results showed that 69% of the samples were contaminated with at least one mycotoxin or metabolite at varying levels. The highest value (max. 152.62 ng/g of ß-ZOL) observed, and the most contaminated livers (42%), were the chicken liver samples. However, pig liver samples presented a high incidence of ZAN (33%) and lamb liver samples presented a high incidence of α-ZOL (40%). The values indicate that there is exposure to these mycotoxins and, although the values are low (ranged to 0.11-152.6 ng/g for α-ZOL and ß-ZOL, respectively), analysis and continuous monitoring are necessary to avoid exceeding the regulatory limits and to control the presence of these mycotoxins in order to protect animal and human health.

Application of an In Vitro Digestion Model for Wheat and Red Beetroot Bread to Assess the Bioaccessibility of Aflatoxin B1, Ochratoxin A and Zearalenone and Betalains.

Nowadays, the bakery industry includes different bioactive ingredients to enrich the nutritional properties of its products, such as betalains from red beetroot (Beta vulgaris). However, cereal products are considered a major route of exposure to many mycotoxins, both individually and in combination, due to their daily consumption, if the cereals used contain these toxins. Only the fraction of the contaminant that is released from the food is bioaccessible and bioavailable to produce toxic effects. Foods with bioactive compounds vary widely in chemical structure and function, and some studies have demonstrated their protective effects against toxics. In this study the bioaccessibility and bioavailability of three legislated mycotoxins (AFB1, OTA and ZEN), individual and combined, in two breads, one with wheat flour and the other with wheat flour enriched with 20% Beta vulgaris, were evaluated. Bioaccessibility of these three mycotoxins from wheat bread and red beet bread enriched individually at 100 ng/g was similar between the breads: 16% and 14% for AFB1, 16% and 17% for OTA and 26% and 22% for ZEN, respectively. Whereas, when mycotoxins were co-present these values varied with a decreasing tendency: 9% and 15% for AFB1, 13% and 9% for OTA, 4% and 25% for ZEN in wheat bread and in red beet bread, respectively. These values reveal that the presence of other components and the co-presence of mycotoxins can affect the final bioavailability; however, it is necessary to assess this process with in vivo studies to complete the studies.

Breast Dense Tissue Segmentation with Noisy Labels: A Hybrid Threshold-Based and Mask-Based Approach.

Breast density assessed from digital mammograms is a known biomarker related to a higher risk of developing breast cancer. Supervised learning algorithms have been implemented to determine this. However, the performance of these algorithms depends on the quality of the ground-truth information, which expert readers usually provide. These expert labels are noisy approximations to the ground truth, as there is both intra- and inter-observer variability among them. Thus, it is crucial to provide a reliable method to measure breast density from mammograms. This paper presents a fully automated method based on deep learning to estimate breast density, including breast detection, pectoral muscle exclusion, and dense tissue segmentation. We propose a novel confusion matrix (CM)-YNet model for the segmentation step. This architecture includes networks to model each radiologist's noisy label and gives the estimated ground-truth segmentation as well as two parameters that allow interaction with a threshold-based labeling tool. A multi-center study involving 1785 women whose "for presentation" mammograms were obtained from 11 different medical facilities was performed. A total of 2496 mammograms were used as the training corpus, and 844 formed the testing corpus. Additionally, we included a totally independent dataset from a different center, composed of 381 women with one image per patient. Each mammogram was labeled independently by two expert radiologists using a threshold-based tool. The implemented CM-Ynet model achieved the highest DICE score averaged over both test datasets (0.82±0.14) when compared to the closest dense-tissue segmentation assessment from both radiologists. The level of concordance between the two radiologists showed a DICE score of 0.76±0.17. An automatic breast density estimator based on deep learning exhibited higher performance when compared with two experienced radiologists. This suggests that modeling each radiologist's label allows for better estimation of the unknown ground-truth segmentation. The advantage of the proposed model is that it also provides the threshold parameters that enable user interaction with a threshold-based tool.

Decision support through risk cost estimation in 30-day hospital unplanned readmission.

Unplanned hospital readmissions mean a significant burden for health systems. Accurately estimating the patient's readmission risk could help to optimise the discharge decision-making process by smartly ordering patients based on a severity score, thus helping to improve the usage of clinical resources. A great number of heterogeneous factors can influence the readmission risk, which makes it highly difficult to be estimated by a human agent. However, this score could be achieved with the help of AI models, acting as aiding tools for decision support systems. In this paper, we propose a machine learning classification and risk stratification approach to assess the readmission problem and provide a decision support system based on estimated patient risk scores.

A deep learning framework to classify breast density with noisy labels regularization.

BACKGROUND AND OBJECTIVE: Breast density assessed from digital mammograms is a biomarker for higher risk of developing breast cancer. Experienced radiologists assess breast density using the Breast Image and Data System (BI-RADS) categories. Supervised learning algorithms have been developed with this objective in mind, however, the performance of these algorithms depends on the quality of the ground-truth information which is usually labeled by expert readers. These labels are noisy approximations of the ground truth, as there is often intra- and inter-reader variability among labels. Thus, it is crucial to provide a reliable method to obtain digital mammograms matching BI-RADS categories. This paper presents RegL (Labels Regularizer), a methodology that includes different image pre-processes to allow both a correct breast segmentation and the enhancement of image quality through an intensity adjustment, thus allowing the use of deep learning to classify the mammograms into BI-RADS categories. The Confusion Matrix (CM) - CNN network used implements an architecture that models each radiologist's noisy label. The final methodology pipeline was determined after comparing the performance of image pre-processes combined with different DL architectures. METHODS: A multi-center study composed of 1395 women whose mammograms were classified into the four BI-RADS categories by three experienced radiologists is presented. A total of 892 mammograms were used as the training corpus, 224 formed the validation corpus, and 279 the test corpus. RESULTS: The combination of five networks implementing the RegL methodology achieved the best results among all the models in the test set. The ensemble model obtained an accuracy of (0.85) and a kappa index of 0.71. CONCLUSIONS: The proposed methodology has a similar performance to the experienced radiologists in the classification of digital mammograms into BI-RADS categories. This suggests that the pre-processing steps and modelling of each radiologist's label allows for a better estimation of the unknown ground truth labels.

Electrical Impedance Tomography Technical Contributions for Detection and 3D Geometric Localization of Breast Tumors: A Systematic Review.

Impedance measuring acquisition systems focused on breast tumor detection, as well as image processing techniques for 3D imaging, are reviewed in this paper in order to define potential opportunity areas for future research. The description of reported works using electrical impedance tomography (EIT)-based techniques and methodologies for 3D bioimpedance imaging of breast tissues with tumors is presented. The review is based on searching and analyzing related works reported in the most important research databases and is structured according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) parameters and statements. Nineteen papers reporting breast tumor detection and location using EIT were systematically selected and analyzed in this review. Clinical trials in the experimental stage did not produce results in most of analyzed proposals (about 80%), wherein statistical criteria comparison was not possible, such as specificity, sensitivity and predictive values. A 3D representation of bioimpedance is a potential tool for medical applications in malignant breast tumors detection being capable to estimate an ap-proximate the tumor volume and geometric location, in contrast with a tumor area computing capacity, but not the tumor extension depth, in a 2D representation.

Echinocandin Drugs Induce Differential Effects in Cytokinesis Progression and Cell Integrity.

Fission yeast contains three essential ß(1,3)-D-glucan synthases (GSs), Bgs1, Bgs3, and Bgs4, with non-overlapping roles in cell integrity and morphogenesis. Only the bgs4+ mutants pbr1-8 and pbr1-6 exhibit resistance to GS inhibitors, even in the presence of the wild-type (WT) sequences of bgs1+ and bgs3+. Thus, Bgs1 and Bgs3 functions seem to be unaffected by those GS inhibitors. To learn more about echinocandins' mechanism of action and resistance, cytokinesis progression and cell death were examined by time-lapse fluorescence microscopy in WT and pbr1-8 cells at the start of treatment with sublethal and lethal concentrations of anidulafungin, caspofungin, and micafungin. In WT, sublethal concentrations of the three drugs caused abundant cell death that was either suppressed (anidulafungin and micafungin) or greatly reduced (caspofungin) in pbr1-8 cells. Interestingly, the lethal concentrations induced differential phenotypes depending on the echinocandin used. Anidulafungin and caspofungin were mostly fungistatic, heavily impairing cytokinesis progression in both WT and pbr1-8. As with sublethal concentrations, lethal concentrations of micafungin were primarily fungicidal in WT cells, causing cell lysis without impairing cytokinesis. The lytic phenotype was suppressed again in pbr1-8 cells. Our results suggest that micafungin always exerts its fungicidal effect by solely inhibiting Bgs4. In contrast, lethal concentrations of anidulafungin and caspofungin cause an early cytokinesis arrest, probably by the combined inhibition of several GSs.

Bias Analysis on Public X-Ray Image Datasets of Pneumonia and COVID-19 Patients.

Chest X-ray images are useful for early COVID-19 diagnosis with the advantage that X-ray devices are already available in health centers and images are obtained immediately. Some datasets containing X-ray images with cases (pneumonia or COVID-19) and controls have been made available to develop machine-learning-based methods to aid in diagnosing the disease. However, these datasets are mainly composed of different sources coming from pre-COVID-19 datasets and COVID-19 datasets. Particularly, we have detected a significant bias in some of the released datasets used to train and test diagnostic systems, which might imply that the results published are optimistic and may overestimate the actual predictive capacity of the techniques proposed. In this article, we analyze the existing bias in some commonly used datasets and propose a series of preliminary steps to carry out before the classic machine learning pipeline in order to detect possible biases, to avoid them if possible and to report results that are more representative of the actual predictive power of the methods under analysis.

Viscoelastic Effects on the Response of Electroelastic Materials.

Electroelastic materials, as for example, 3M VHB 4910, are attracting attention as actuators or generators in some developments and applications. This is due to their capacity of being deformed when submitted to an electric field. Some models of their actuation are available, but recently, viscoelastic models have been proposed to give an account of the dissipative behaviour of these materials. Their response to an external mechanical or electrical force field implies a relaxation process towards a new state of thermodynamic equilibrium, which can be described by a relaxation time. However, it is well known that viscoelastic and dielectric materials, as for example, polymers, exhibit a distribution of relaxation times instead of a single relaxation time. In the present approach, a continuous distribution of relaxation times is proposed via the introduction of fractional derivatives of the stress and strain, which gives a better account of the material behaviour. The application of fractional derivatives is described and a comparison with former results is made. Then, a double generalisation is carried out: the first one is referred to the viscoelastic or dielectric models and is addressed to obtain a nonsymmetric spectrum of relaxation times, and the second one is the adoption of the more realistic Mooney-Rivlin equation for the stress-strain relationship of the elastomeric material. A modified Mooney-Rivlin model for the free energy density of a hyperelastic material, VHB 4910 has been used based on experimental results of previous authors. This last proposal ensures the appearance of the bifurcation phenomena which is analysed for equibiaxial dead loads; time-dependent bifurcation phenomena are predicted by the extended Mooney-Rivlin equations.

Analysis and application of a suite of recombinant endo-ß(1,3)-D-glucanases for studying fungal cell walls.

BACKGROUND: The fungal cell wall is an essential and robust external structure that protects the cell from the environment. It is mainly composed of polysaccharides with different functions, some of which are necessary for cell integrity. Thus, the process of fractionation and analysis of cell wall polysaccharides is useful for studying the function and relevance of each polysaccharide, as well as for developing a variety of practical and commercial applications. This method can be used to study the mechanisms that regulate cell morphogenesis and integrity, giving rise to information that could be applied in the design of new antifungal drugs. Nonetheless, for this method to be reliable, the availability of trustworthy commercial recombinant cell wall degrading enzymes with non-contaminating activities is vital. RESULTS: Here we examined the efficiency and reproducibility of 12 recombinant endo-ß(1,3)-D-glucanases for specifically degrading the cell wall ß(1,3)-D-glucan by using a fast and reliable protocol of fractionation and analysis of the fission yeast cell wall. This protocol combines enzymatic and chemical degradation to fractionate the cell wall into the four main polymers: galactomannoproteins, α-glucan, ß(1,3)-D-glucan and ß(1,6)-D-glucan. We found that the GH16 endo-ß(1,3)-D-glucanase PfLam16A from Pyrococcus furiosus was able to completely and reproducibly degrade ß(1,3)-D-glucan without causing the release of other polymers. The cell wall degradation caused by PfLam16A was similar to that of Quantazyme, a recombinant endo-ß(1,3)-D-glucanase no longer commercially available. Moreover, other recombinant ß(1,3)-D-glucanases caused either incomplete or excessive degradation, suggesting deficient access to the substrate or release of other polysaccharides. CONCLUSIONS: The discovery of a reliable and efficient recombinant endo-ß(1,3)-D-glucanase, capable of replacing the previously mentioned enzyme, will be useful for carrying out studies requiring the digestion of the fungal cell wall ß(1,3)-D-glucan. This new commercial endo-ß(1,3)-D-glucanase will allow the study of the cell wall composition under different conditions, along the cell cycle, in response to environmental changes or in cell wall mutants. Furthermore, this enzyme will also be greatly valuable for other practical and commercial applications such as genome research, chromosomes extraction, cell transformation, protoplast formation, cell fusion, cell disruption, industrial processes and studies of new antifungals that specifically target cell wall synthesis.

Probabilistic Evaluation of 3D Surfaces Using Statistical Shape Models (SSM).

Inspecting a 3D object which shape has elastic manufacturing tolerances in order to find defects is a challenging and time-consuming task. This task usually involves humans, either in the specification stage followed by some automatic measurements, or in other points along the process. Even when a detailed inspection is performed, the measurements are limited to a few dimensions instead of a complete examination of the object. In this work, a probabilistic method to evaluate 3D surfaces is presented. This algorithm relies on a training stage to learn the shape of the object building a statistical shape model. Making use of this model, any inspected object can be evaluated obtaining a probability that the whole object or any of its dimensions are compatible with the model, thus allowing to easily find defective objects. Results in simulated and real environments are presented and compared to two different alternatives.

Anesthesia assessment in the era of competences: state of the art / La evaluación en anestesia en la era de las competencias: estado del arte

Abstract Introduction: Anesthesiology requires procedure fulfillment, problem, and real-time crisis resolution, problem, and complications forecast, among others; therefore, the evaluation of its learning should center around how students achieve competence rather than solely focusing on knowledge acquisition. Literature shows that despite the existence of numerous evaluation strategies, these are still underrated in most cases due to unawareness. Objective: The present article aims to explain the process of competency-based anesthesiology assessment, in addition to suggesting a brief description of the learning domains evaluated, theories of knowledge, instruments, and assessment systems in the area; and finally, to show some of the most relevant results regarding assessment systems in Colombia. Methodology: The results obtained in "Characteristics of the evaluation systems used by anesthesiology residency programs in Colombia" showed a certain degree of unawareness by stakeholders in the educational process, a fact that motivated the publishing of this discussion around the topic of competency-based assessment in anesthesiology. Following a bibliography search with the keywords through PubMed, OVID, ERIC, DIALNET, and REDALYC, 110 articles were reviewed and 75 were established as relevant for the research's theoretical framework. Results and conclusion: Anesthesiology assessment should be conceived from the competency's multidimensionality; it must be longitudinal and focused on the learning objectives.

Resumen Introducción: La anestesiología requiere la realización de procedimientos, resolución de problemas y crisis en tiempo real, previsión de problemas y complicaciones, entre otros, por lo tanto, la evaluación de su aprendizaje debería centrarse en cómo el estudiante alcanza la competencia y no solo en la adquisición de conocimientos. La literatura muestra que, a pesar de existir numerosas estrategias de evaluación, estas continúan siendo subvaloradas en muchos casos por desconocimiento. Objetivo: Este artículo pretende dar a conocer el proceso de evaluación en la anestesiología desde la competencia, además de sugerir una breve descripción de los dominios y teorías de aprendizaje, instrumentos y sistemas de evaluación en esta área y, finalmente, mostrar algunos de los resultados más relevantes sobre los sistemas de evaluación en Colombia. Metodología: Tras una búsqueda bibliográfica en PubMed, OVID, ERIC, DIALNET, REDALYC, con las palabras clave, se revisaron 110 artículos de los cuales 75 fueron considerados relevantes para elaborar el marco teórico de la investigación. Resultados y conclusiones: La evaluación en anestesiología debe ser concebida desde la multidimensionalidad de la competencia, ser longitudinal y enfocada en los objetivos de aprendizaje.

A deep learning system to obtain the optimal parameters for a threshold-based breast and dense tissue segmentation.

BACKGROUND AND OBJECTIVE: Breast cancer is the most frequent cancer in women. The Spanish healthcare network established population-based screening programs in all Autonomous Communities, where mammograms of asymptomatic women are taken with early diagnosis purposes. Breast density assessed from digital mammograms is a biomarker known to be related to a higher risk to develop breast cancer.It is thus crucial to provide a reliable method to measure breast density from mammograms. Furthermore the complete automation of this segmentation process is becoming fundamental as the amount of mammograms increases every day. Important challenges are related with the differences in images from different devices and the lack of an objective gold standard.This paper presents a fully automated framework based on deep learning to estimate the breast density. The framework covers breast detection, pectoral muscle exclusion, and fibroglandular tissue segmentation. METHODS: A multi-center study, composed of 1785 women whose "for presentation" mammograms were segmented by two experienced radiologists. A total of 4992 of the 6680 mammograms were used as training corpus and the remaining (1688) formed the test corpus. This paper presents a histogram normalization step that smoothed the difference between acquisition, a regression architecture that learned segmentation parameters as intrinsic image features and a loss function based on the DICE score. RESULTS: The results obtained indicate that the level of concordance (DICE score) reached by the two radiologists (0.77) was also achieved by the automated framework when it was compared to the closest breast segmentation from the radiologists. For the acquired with the highest quality device, the DICE score per acquisition device reached 0.84, while the concordance between radiologists was 0.76. CONCLUSIONS: An automatic breast density estimator based on deep learning exhibits similar performance when compared with two experienced radiologists. It suggests that this system could be used to support radiologists to ease its work.

Effect of the early use of antivirals on the COVID-19 pandemic. A computational network modeling approach.

It seems that we are far from controlling COVID-19 pandemics, and, consequently, returning to a fully normal life. Until an effective vaccine is found, safety measures as the use of face masks, social distancing, washing hands regularly, etc., have to be taken. Also, the use of appropriate antivirals in order to alleviate the symptoms, to control the severity of the illness and to prevent the transmission, could be a good option that we study in this work. In this paper, we propose a computational random network model to study the transmission dynamics of COVID-19 in Spain. Once the model has been calibrated and validated, we use it to simulate several scenarios where effective antivirals are available. The results show how the early use of antivirals may significantly reduce the incidence of COVID-19 and may avoid a new collapse of the health system.

High-Profile VRU Detection on Resource-Constrained Hardware Using YOLOv3/v4 on BDD100K.

Vulnerable Road User (VRU) detection is a major application of object detection with the aim of helping reduce accidents in advanced driver-assistance systems and enabling the development of autonomous vehicles. Due to intrinsic complexity present in computer vision and to limitations in processing capacity and bandwidth, this task has not been completely solved nowadays. For these reasons, the well established YOLOv3 net and the new YOLOv4 one are assessed by training them on a huge, recent on-road image dataset (BDD100K), both for VRU and full on-road classes, with a great improvement in terms of detection quality when compared to their MS-COCO-trained generic correspondent models from the authors but with negligible costs in forward pass time. Additionally, some models were retrained when replacing the original Leaky ReLU convolutional activation functions from original YOLO implementation with two cutting-edge activation functions: the self-regularized non-monotonic function (MISH) and its self-gated counterpart (SWISH), with significant improvements with respect to the original activation function detection performance. Additionally, some trials were carried out including recent data augmentation techniques (mosaic and cutmix) and some grid size configurations, with cumulative improvements over the previous results, comprising different performance-throughput trade-offs.