Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) has been recognized as an effective tool for Breast Cancer (BC) diagnosis. Automatic BC analysis from DCE-MRI depends on features extracted particularly from lesions, hence, lesions need to be accurately segmented as a prior step. Due to the time and experience required to manually segment lesions in 4D DCE-MRI, automating this task is expected to reduce the workload, reduce observer variability and improve diagnostic accuracy. In this paper we propose an automated method for breast lesion segmentation from DCE-MRI based on a U-Net framework. The contributions of this work are the proposal of a modified U-Net architecture and the analysis of the input DCE information. In that sense, we propose the use of an ensemble method combining three U-Net models, each using a different input combination, outperforming all individual methods and other existing approaches. For evaluation, we use a subset of 46 cases from the TCGA-BRCA dataset, a challenging and publicly available dataset not reported to date for this task. Due to the incomplete annotations provided, we complement them with the help of a radiologist in order to include secondary lesions that were not originally segmented. The proposed ensemble method obtains a mean Dice Similarity Coefficient (DSC) of 0.680 (0.802 for main lesions) which outperforms state-of-the art methods using the same dataset, demonstrating the effectiveness of our method considering the complexity of the dataset.

BACKGROUND: CT findings of lung cancer and tuberculosis are sometimes similar, potentially leading to misdiagnosis. This study aims to combine deep learning and content-based image retrieval (CBIR) to distinguish lung cancer (LC) from nodular/mass atypical tuberculosis (NMTB) in CT images. METHODS: This study proposes CBIR with a convolutional Siamese neural network (CBIR-CSNN). First, the lesion patches are cropped out to compose LC and NMTB datasets and the pairs of two arbitrary patches form a patch-pair dataset. Second, this patch-pair dataset is utilized to train a CSNN. Third, a test patch is treated as a query. The distance between this query and 20 patches in both datasets is calculated using the trained CSNN. The patches closest to the query are used to give the final prediction by majority voting. One dataset of 719 patients is used to train and test the CBIR-CSNN. Another external dataset with 30 patients is employed to verify CBIR-CSNN. RESULTS: The CBIR-CSNN achieves excellent performance at the patch level with an mAP (Mean Average Precision) of 0.953, an accuracy of 0.947, and an area under the curve (AUC) of 0.970. At the patient level, the CBIR-CSNN correctly predicted all labels. In the external dataset, the CBIR-CSNN has an accuracy of 0.802 and AUC of 0.858 at the patch level, and 0.833 and 0.902 at the patient level. CONCLUSIONS: This CBIR-CSNN can accurately and automatically distinguish LC from NMTB using CT images. CBIR-CSNN has excellent representation capability, compatibility with few-shot learning, and visual explainability.

Fat accumulation in the liver cells can increase the risk of cardiac complications and cardiovascular disease mortality. Therefore, a way to quickly and accurately detect hepatic steatosis is critically important. However, current methods, e.g., liver biopsy, magnetic resonance imaging, and computerized tomography scan, are subject to high cost and/or medical complications. In this paper, we propose a deep neural network to estimate the degree of hepatic steatosis (low, mid, high) using only body shapes. The proposed network adopts dilated residual network blocks to extract refined features of input body shape maps by expanding the receptive field. Furthermore, to classify the degree of steatosis more accurately, we create a hybrid of the center loss and cross entropy loss to compact intra-class variations and separate inter-class differences. We performed extensive tests on the public medical dataset with various network parameters. Our experimental results show that the proposed network achieves a total accuracy of over 82% and offers an accurate and accessible assessment for hepatic steatosis.

PURPOSE: To predict acute kidney injury (AKI) in a large intensive care unit (ICU) database. MATERIALS AND METHODS: A total of 30,020 ICU admissions with 17,222 AKI episodes were extracted from the Medical Information Mart from Intensive Care (MIMIC)-III database. These were randomly divided into a training set and an independent testing set in a ratio of 4:1. Data pertaining to demographics, admission information, vital signs, laboratory tests, critical illness scores, medications, comorbidities, and intervention measures were collected. Logistic regression, random forest, LightGBM, XGBoost, and an ensemble model was used for early prediction of AKI occurrence and important feature extraction. The SHAP analysis was adopted to reveal the impact of prediction for each feature. RESULTS: The ensemble model had the best overall performance for predicting AKI before 24 h, 48 h and 72 h. The F1 values were 0.915, 0.893, and 0.878, respectively. AUCs were 0.923, 0.903, and 0.895, respectively. CONCLUSIONS: Based on readily available electronic medical record (EMR) data, gradient boosting decision tree models are highly accurate at early AKI prediction in critically ill patients.

The outbreak of a new coronavirus (SARS-CoV-2) was first identified in Wuhan, People's Republic of China, in 2019, which has led to a severe, life-threatening form of pneumonia (COVID-19). Research scientists all around the world have been trying to find small molecule drugs to treat COVID-19. In the present study, a conserved macrodomain, ADP Ribose phosphatase (ADRP), of a critical non-structural protein (Nsp3) in all coronaviruses was probed using large-scale Molecular Dynamics (MD) simulations to identify novel inhibitors. In our virtual screening workflow, the recently-solved X-ray complex structure, 6W6Y, with a substrate-mimics was used to screen 17 million ZINC15 compounds using drug property filters and Glide docking scores. The top twenty output compounds each underwent 200 ns MD simulations (i.e. 20 × 200 ns) to validate their individual stability as potential inhibitors. Eight out of the twenty compounds showed stable binding modes in the MD simulations, as well as favorable drug properties from our predctions. Therefore, our computational data suggest that the resulting top eight out of twenty compounds could potentially be novel inhibitors to ADRP of SARS-CoV-2.

Lung cancer causes more than one million deaths worldwide each year. Averages of 5-year survival rate of patients with Non-small cell lung cancer (NSCLC), which is the most common type of lung cancer, is 15%. Computer-Aided Detection (CAD) is a very important tool for identifying lung lesions in medical imaging. In general, the process line of a CAD system can be divided into four main stages: preprocessing, localization, feature extraction, and classification. As segmentation is required for localization in computer vision and medical image analysis, this step has become a major and challenging problem, and much research has been done on new segmentation techniques. In recent years, interest in model-based segmentation methods has increased, and the reason for this is even if some object information is lost, such gaps can be filled by using the previous information in the model. This paper proposed Texture Appearance Model (TAM), which is a new model-based method and segments all types of nodule areas accurately and efficiently, including juxta-pleural nodules, without separating the lung from the surrounding area in a CT scan of the lung. In this method, Texture Representation of Image (TRI) is obtained using tissue texture feature extraction and feature selection algorithms. The proposed method has been evaluated in 85 nodules of the dataset, received from the Iranian hospital, in which the ground-truth annotation by physicians and CT imaging data were provided. The results show that the proposed algorithm has an encouraging performance for distinguishing different types of nodules, including pleural, cavity and non-solid nodules, achieving an average dice similarity coefficient (DSC) of 84.75%.

BACKGROUND AND OBJECTIVE: Hypertensive intracerebral hemorrhage is characterized by a high rate of morbidity, mortality, disability and recurrence. Neuroendoscopy has been utilized for treatment as an advanced technology. However, traditional neuroendoscopy allows professionals to see only tissue surfaces, and the field of vision is limited, which cannot provide spatial guidance. In this study, an AR-based neuroendoscopic navigation system is proposed to assist surgeons in locating and clearing hematoma. METHODS: The neuroendoscope can be registered through the vector closed loop algorithm. The single-shot method is designed to register medical images with patients precisely. Real-time AR is realized based on video stream fusion. Dual-mode AR navigation is proposed to provide comprehensive guidance from catheter implantation to hematoma removal. A series of experiments is designed to validate the accuracy and significance of this system. RESULTS: The average root mean square error of the registration between medical images and patients is 0.784 mm, and the variance is 0.1426 mm. The pixel mismatching degrees are less than 1% in different AR modes. In catheter implantation experiments, the average error of distance is 1.28 mm, and the variance is 0.43 mm, while the average error of angles is 1.34°, and the variance is 0.45°. Comparative experiments are also conducted to evaluate the feasibility of this system. CONCLUSION: This system can provide stereo images with depth information fused with patients to guide surgeons to locate targets and remove hematoma. It has been validated to have high accuracy and feasibility.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused a global pandemic of coronavirus disease 2019 (COVID-19) since December 2019 that has led to more than 160 million confirmed cases, including 3.3 million deaths. To understand the mechanism by which SARS-CoV-2 invades human cells and reveal organ-specific susceptible cell types for COVID-19, we conducted comprehensive bioinformatic analysis using public single-cell RNA sequencing datasets. Utilizing the expression information of six confirmed COVID-19 receptors (ACE2, TMPRSS2, NRP1, AXL, FURIN and CTSL), we demonstrated that macrophages are the most likely cells that may be associated with SARS-CoV-2 pathogenesis in lung. Besides the widely reported 'chemokine storm', we identified ribosome related pathways that may also be potential therapeutic target for COVID-19 lung infection patients. Moreover, cell-cell communication analysis and trajectory analysis revealed that M1-like macrophages showed the highest relation to severe COVID-19 patients. And we also demonstrated that up-regulation of chemokine pathways generally lead to severe symptoms, while down-regulation of ribosome and RNA activity related pathways are more likely to be mild. Other organ-specific susceptible cell type analyses could also provide potential targets for COVID-19 therapy. This work can provide clues for understanding the pathogenesis of COVID-19 and contribute to understanding the mechanism by which SARS-CoV-2 invades human cells.

Venetoclax is a BH3 (BCL-2 Homology 3) mimetic used to treat leukemia and lymphoma by inhibiting the anti-apoptotic BCL-2 protein thereby promoting apoptosis of cancerous cells. Acquired resistance to Venetoclax via specific variants in BCL-2 is a major problem for the successful treatment of cancer patients. Replica exchange molecular dynamics (REMD) simulations combined with machine learning were used to define the average structure of variants in aqueous solution to predict changes in drug and ligand binding in BCL-2 variants. The variant structures all show shifts in residue positions that occlude the binding groove, and these are the primary contributors to drug resistance. Correspondingly, we established a method that can predict the severity of a variant as measured by the inhibitory constant (Ki) of Venetoclax by measuring the structure deviations to the binding cleft. In addition, we also applied machine learning to the phi and psi angles of the amino acid backbone to the ensemble of conformations that demonstrated a generalizable method for drug resistant predictions of BCL-2 proteins that elucidates changes where detailed understanding of the structure-function relationship is less clear.

To design a flapping-wing micro air vehicle (FWMAV), the hovering flight action of a beetle species (Protaetia brevitarsis) was captured, and various parameters, such as the hindwing flapping frequency, flapping amplitude, angle of attack, rotation angle, and stroke plane angle, were obtained. The wing tip trajectories of the hindwings were recorded and analyzed, and the flapping kinematics were assessed. Based on the wing tip trajectory functions, bioinspired wings and a linkage mechanism flapping system were designed. The critical parameters for the aerodynamic characteristics were investigated and optimized by means of wind tunnel tests, and the artificial flapping system with the best wing parameters was compared with the natural beetle. This work provides insight into how natural flyers execute flight by experimentally duplicating beetle hindwing kinematics and paves the way for the future development of beetle-mimicking FWMAVs.

BACKGROUND: Accessibility of labelled datasets is often a key limitation for the application of Machine Learning in clinical research. A novel semi-automated weak-labelling approach based on unsupervised clustering was developed to classify a large dataset of microneurography signals and subsequently used to train a Neural Network to reproduce the labelling process. METHODS: Clusters of microneurography signals were created with k-means and then labelled in terms of the validity of the signals contained in each cluster. Only purely positive or negative clusters were labelled, whereas clusters with mixed content were passed on to the next iteration of the algorithm to undergo another cycle of unsupervised clustering and labelling of the clusters. After several iterations of this process, only pure labelled clusters remained which were used to train a Deep Neural Network. RESULTS: Overall, 334,548 individual signal peaks form the integrated data were extracted and more than 99.99% of the data was labelled in six iterations of this novel application of weak labelling with the help of a domain expert. A Deep Neural Network trained based on this dataset achieved consistent accuracies above 95%. DISCUSSION: Data extraction and the novel iterative approach of labelling unsupervised clusters enabled creation of a large, labelled dataset combining unsupervised learning and expert ratings of signal-peaks on cluster basis in a time effective manner. Further research is needed to validate the methodology and employ it on other types of physiologic data for which it may enable efficient generation of large labelled datasets.

Despite impressive developments in deep convolutional neural networks for medical imaging, the paradigm of supervised learning requires numerous annotations in training to avoid overfitting. In clinical cases, massive semantic annotations are difficult to acquire where biomedical expert knowledge is required. Moreover, it is common when only a few annotated classes are available. In this study, we proposed a new approach to few-shot medical image segmentation, which enables a segmentation model to quickly generalize to an unseen class with few training images. We constructed a few-shot image segmentation mechanism using a deep convolutional network trained episodically. Motivated by the spatial consistency and regularity in medical images, we developed an efficient global correlation module to model the correlation between a support and query image and incorporate it into the deep network. We enhanced the discrimination ability of the deep embedding scheme to encourage clustering of feature domains belonging to the same class while keeping feature domains of different organs far apart. We experimented using anatomical abdomen images from both CT and MRI modalities.

Emotion recognition is a vital but challenging step in creating passive brain-computer interface applications. In recent years, many studies on electroencephalogram (EEG)-based emotion recognition have been conducted. Ensemble learning has been widely used in emotion recognition because of its superior accuracy and generalization. In this study, we proposed a novel ensemble learning method based on multiple objective particle swarm optimization for subject-independent EEG-based emotion recognition. First, we used a 4 s sliding time window with a 2 s overlap to extract 13 different features from EEG signals and construct a feature vector. Then, we employed L1 regularization to select effective features. Second, a model selection method was applied to choose the optimal basic analysis submodels. Afterward, we proposed an ensemble operator that converts the classification results of a single model from discrete values to continuous values to better characterize the classification results. Subsequently, multiple objective particle swarm optimization was adopted to confirm the optimal parameters of the ensemble learning model. Finally, we conducted extensive experiments on two public datasets: DEAP and SEED. Considering the generalization of the model, we applied leave-one-subject-out cross-validation to evaluate the performance of the model. The experimental results demonstrate that the proposed method achieves a better recognition performance than single methods, commonly used ensemble learning methods, and state-of-the-art methods. The average accuracies for arousal and valence are 65.70% and 64.22%, respectively, on the DEAP database, and the average accuracy on the SEED database is 84.44%.

BACKGROUND: Carotid endarterectomy (CEA) remains the first-line treatment option of symptomatic and asymptomatic carotid stenosis, while stenting (CAS) is reserved for selected patients at high surgical risk. Here, we compare the vascular remodeling process in CEA- and CAS-treated patients with respect to morphological and hemodynamic features, because of their possible engagement in carotid atherosclerosis. METHODS: Twelve (12) patients were included, half with patched CEA and half with CAS. Pre- and post-operative 3D image-based models of the carotid bifurcation were anatomically characterized in terms of flare, tortuosity, and curvature. Individual computational fluid dynamics simulations allowed to quantify the postoperative hemodynamic milieu in terms of (1) wall shear stress and (2) helical flow. RESULTS: Carotid flare increased in all cases, but a more marked increase emerged after CEA compared to CAS. Tortuosity and curvature increased after CEA but decreased after CAS. CEA patients presented with significantly higher postoperative tortuosity than CAS patients. CEA was associated with a worse (non-statistically significant) score in all flow disturbance indicators vs. CAS. CONCLUSION: The increased flare and tortuosity of the carotid bifurcation after CEA vs. CAS is a marked difference in the vascular remodeling process between the two modalities. CAS seems to induce a less pro-restenosis hemodynamic environment compared to CEA. The emerged differences stimulate further analysis on a larger cohort with long-term outcomes, to shed light on the clinical impact of the observations.

Lead position is an important factor in determining response to Cardiac Resynchronization Therapy (CRT) in dyssynchronous heart failure (HF) patients. Multipoint pacing (MPP) enables pacing from multiple electrodes within the same lead, improving the potential outcome for patients. Virtual quadripolar lead designs were evaluated by simulating pacing from all combinations of 1 and 2 electrodes along the lead in each virtual patient from cohorts of HF (n = 24) and simulated reverse remodelled (RR, n = 20) patients. Electrical synchrony was assessed by the time 90% of the ventricular myocardium is activated (AT090). Optimal 1 and 2 electrode pacing configurations for AT090 were combined to identify the 4-electrode lead design that maximised benefits across all patients. LV pacing in the HF cohort in all possible single and double electrode locations reduced AT090 by 14.48 ± 5.01 ms (11.92 ± 3.51%). The major determinant of reduction in activation time was patient anatomy. Pacing with a single optimal lead design reduced AT090 more in the HF cohort than the RR cohort (12.68 ± 3.29% vs 10.81 ± 2.34%). Pacing with a single combined HF and RR population-optimised lead design achieves electrical resynchronization with near equivalence to personalised lead designs both in HF and RR anatomies. These findings suggest that although lead configurations have to be tailored to each patient, a single optimal lead design is sufficient to obtain near-optimal results across most patients. This study shows the potential of virtual clinical trials as tools to compare existing and explore new lead designs.

PURPOSE: Since Generative Adversarial Network (GAN) was introduced into the field of deep learning in 2014, it has received extensive attention from academia and industry, and a lot of high-quality papers have been published. GAN effectively improves the accuracy of medical image segmentation because of its good generating ability and capability to capture data distribution. This paper introduces the origin, working principle, and extended variant of GAN, and it reviews the latest development of GAN-based medical image segmentation methods. METHOD: To find the papers, we searched on Google Scholar and PubMed with the keywords like "segmentation", "medical image", and "GAN (or generative adversarial network)". Also, additional searches were performed on Semantic Scholar, Springer, arXiv, and the top conferences in computer science with the above keywords related to GAN. RESULTS: We reviewed more than 120 GAN-based architectures for medical image segmentation that were published before September 2021. We categorized and summarized these papers according to the segmentation regions, imaging modality, and classification methods. Besides, we discussed the advantages, challenges, and future research directions of GAN in medical image segmentation. CONCLUSIONS: We discussed in detail the recent papers on medical image segmentation using GAN. The application of GAN and its extended variants has effectively improved the accuracy of medical image segmentation. Obtaining the recognition of clinicians and patients and overcoming the instability, low repeatability, and uninterpretability of GAN will be an important research direction in the future.

Munro's Microabscess (MM) is the diagnostic hallmark of psoriasis. Neutrophil detection in the Stratum Corneum (SC) of the skin epidermis is an integral part of MM detection in skin biopsy. The microscopic inspection of skin biopsy is a tedious task and staining variations in skin histopathology often hinder human performance to differentiate neutrophils from skin keratinocytes. Motivated from this, we propose a computational framework that can assist human experts and reduce potential errors in diagnosis. The framework first segments the SC layer, and multiple patches are sampled from the segmented regions which are classified to detect neutrophils. Both UNet and CapsNet are used for segmentation and classification. Experiments show that of the two choices, CapsNet, owing to its robustness towards better hierarchical object representation and localisation ability, appears as a better candidate for both segmentation and classification tasks and hence, we termed our framework as MICaps. The training algorithm explores both minimisation of Dice Loss and Focal Loss and makes a comparative study between the two. The proposed framework is validated with our in-house dataset consisting of 290 skin biopsy images. Two different experiments are considered. Under the first protocol, only 3-fold cross-validation is done to directly compare the current results with the state-of-the-art ones. Next, the performance of the system on a held-out data set is reported. The experimental results show that MICaps improves the state-of-the-art diagnosis performance by 3.27% (maximum) and reduces the number of model parameters by 50%.

Major depressive disorder (MDD) is a common mental illness characterized by persistent feeling of depressed mood and loss of interest. It would cause, in a severe case, suicide behaviors. In clinical settings, automatic MDD detection is mainly based on electroencephalogram (EEG) signals with supervised learning techniques. However, supervised-based MDD detection methods encounter two ineviTable bottlenecks: firstly, such methods rely heavily on an EEG training dataset with MDD labels annotated by a physical therapist, leading to subjectivity and high cost; secondly, most of EEG signals are unlabeled in a real scenario. In this paper, a novel semisupervised-based MDD detection method named MDD-TSVM is presented. Specifically, the MDD-TSVM utilizes the semisupervised method of transductive support vector machine (TSVM) as its backbone, further dividing the unlabeled penalty item of the TSVM objective function into two pseudo-labeled penalty items with or without MDD. By such improvement, the MDD-SVM can make full use of labeled and unlabeled datasets as well as alleviate the class imbalance problem. Experiment results showed that our proposed MDD-TSVM achieved F1 score of 0.85 ± 0.05 and accuracy of 0.89 ± 0.03 on identifying MDD patients, which is superior to the state-of-the-art methods.

One of the most characteristic signs of Parkinson's disease (PD) is hand tremor. The MDS-UPDRS scale evaluates different aspects of the disease. The tremor score is a part of the MDS-UPDRS scale, which provides instructions for rating it, by observation, with an integer from 0 to 4. Nevertheless, this form of assessment is subjective and dependent on visual acuity, clinical judgment, and even the mood of the individual examiner. On the other hand, in many cases, existing computational models proposed to resolve the disadvantages of the MDS-UPDRS scale may have uncertainty in differentiating a category of a slight Parkinson tremor from voluntary movements. In this study, 554 measurements from Parkinson's patients, and 60 measurements from healthy subjects, were recorded with inertial sensors placed on the back of each hand. Five biomechanical indicators characterised the hand tremor. With these indicators, the three fuzzy inference models proposed can differentiate, in the first instance, the presence of postural or resting tremor from a normal movement of the hand, and if detected, to determine its severity. The fuzzy inference models allowed following the criteria of the MDS-UPDRS scale, providing an evaluation with an accuracy of two decimal digits and which, due to its simplicity, can be implemented in clinical environments. The assessments of three experts validated the computer model.

Vaccination remains the most effective strategy for preventing and controlling infectious diseases. Numerous conventional vaccines, especially live attenuated, inactivated (killed) microorganisms and subunit vaccines, lead to an effective induction of protective immune responses, mainly antibody-mediated responses against pathogens. However, it has become known that a wide range of highly dangerous pathogens are uncontrollable via conventional vaccination strategies. Recent advances in molecular biology, immunology, genetics, biochemistry, and bioinformatics have provided new prospects for vaccine development. As a result of these advances, several new strategies for vaccine design, development, and production have appeared. These strategies show advantages over conventional vaccines. In this review, we discuss some of the major novel approaches, including recombinant protein vaccines, live recombinant viral and bacterial vectors, DNA and RNA vaccines, reverse vaccinology and reverse genetics approaches. Moreover, we have described the recent progresses on computational tools and immunoinformatics approaches for identifying, designing, and developing new candidate vaccines.

Although biofilm-specific antibiotic susceptibility assays are available, they are time-consuming and resource-intensive, and hence they are not usually performed in clinical settings. Herein, we introduce a machine learning-based predictive modeling approach that uses routinely available and easily accessible data to qualitatively predict in vitro antibiofilm activity of antibiotics with relatively high accuracy. Three optimized models based on logistic regression, decision tree, and random forest algorithms were successfully developed in this study using data manually collected from published literature. In these models, independent variables that serve as significant predictors of antibiofilm activity are minimum inhibitory concentration, bacterial Gram type, biofilm formation method, in addition to antibiotic's mechanism of action, molecular weight, and pKa. The cross-validation method showed that the optimized models exhibit prediction accuracy of 67% ± 6.1% for the logistic regression model, 73% ± 5.8% for the decision tree model, and 74% ± 5% for the random forest model. However, the one-way ANOVA test revealed that the difference in prediction accuracy between the 3 models is not statistically significant, and hence they can be considered to have comparable performance. The presented modeling approach can serve as an alternative to the resource-intensive biofilm assays to rapidly and properly manage biofilm-associated infections, especially in resource-limited clinical settings.

Despite remarkable progress in disease diagnosis and treatment, coronary heart disease (CHD) remains the number one leading cause of death worldwide. Many practical challenges still faced in clinical settings necessitates the pursuit of omics studies to identify alternative/orthogonal biomarkers, as well as to discover novel insights into disease mechanisms. Albeit relatively nascent as compared to the omics frontrunners (genomics, transcriptomics, and proteomics), omics beyond the central dogma (OBCD; e.g., metabolomics, lipidomics, glycomics, and metallomics) have undeniable contributions and prospects in CHD research. In this bibliometric study, we characterised the global trends in publication/citation outputs, collaborations, and research hotspots concerning OBCD-CHD, with a focus on the more prolific fields of metabolomics and lipidomics. As for glycomics and metallomics, there were insufficient publication records on their applications in CHD research for quantitative bibliometrics analysis. Thus, we reviewed their applications in health/disease research in general, discussed and justified their potential in CHD research, and suggested important/promising research avenues. By summarising evidence obtained both quantitatively and qualitatively, this study offers a first and comprehensive picture of OBCD applications in CHD, facilitating the establishment of future research directions.

Diabetic foot ulcer (DFU) is a major complication of diabetes and can lead to lower limb amputation if not treated early and properly. In addition to the traditional clinical approaches, in recent years, research on automation using computer vision and machine learning methods plays an important role in DFU classification, achieving promising successes. The most recent automatic approaches to DFU classification are based on convolutional neural networks (CNNs), using solely RGB images as input. In this paper, we present a CNN-based DFU classification method in which we showed that feeding an appropriate feature (texture information) to the CNN model provides a complementary performance to the standard RGB-based deep models of the DFU classification task, and better performance can be obtained if both RGB images and their texture features are combined and used as input to the CNN. To this end, the proposed method consists of two main stages. The first stage extracts texture information from the RGB image using the mapped binary patterns technique. The obtained mapped image is used to aid the second stage in recognizing DFU as it contains texture information of ulcer. The stack of RGB and mapped binary patterns images are fed to the CNN as a tensor input or as a fused image, which is a linear combination of RGB and mapped binary patterns images. The performance of the proposed approach was evaluated using two recently published DFU datasets: the Part-A dataset of healthy and unhealthy (DFU) cases [17] and Part-B dataset of ischaemia and infection cases [18]. The results showed that the proposed methods provided better performance than the state-of-the-art CNN-based methods with 0.981% (AUC) and 0.952% (F-Measure) on the Part-A dataset, 0.995% (AUC) and 0.990% (F-measure) for the Part-B ischaemia dataset, and 0.820% (AUC) and 0.744% (F-measure) on the Part-B infection dataset.

Automatic pollen images recognition is crucial for pollinosis symptoms prevention and treatment. The problem of pollen recognition can be efficiently solved using deep learning, however neural networks require tens of thousands of images to generalize. At the same time, the existing open pollen images datasets are very small. In this paper, we present a novel open pollen dataset annotated for both detection and classification tasks. Based on our dataset we study learning from a small data using different state-of-the-art approaches. For the detection task we propose to use our new Bayesian RetinaNet network, which models aleatoric uncertainty. We compare it with the baseline RetinaNet and demonstrate that our model allows for higher detection precision. For the classification task we compare the impact of pre-training on the synthetic images from generative adversarial networks (GANs) and metric-based few-shot learning. Namely, we pre-trained our small convolutional neural network and Siamese neural network classifiers on synthetic pollen images generated by two GANs: StyleGAN and Self-attention GAN. The best classifier is the convolutional neural network pre-trained on StyleGAN images. Our best models achieved 96.3% of mean average precision for the detection task and 97.7% of F1 measure for the classification task on 13 pollen plant species. We have implemented the best models in our pollen recognition web service, which is available for palynologists on request.

Quantitative assessment of the complex hemodynamic environment in type B aortic dissection (TBAD) through computational fluid dynamics (CFD) simulations can provide detailed insights into the disease and its progression. As imaging and computational technologies have advanced, methodologies have been developed to increase the accuracy and physiological relevance of CFD simulations. This study presents a patient-specific workflow to simulate blood flow in TBAD, utilising the maximum amount of in vivo data available in the form of CT images, 4D-flow MRI and invasive Doppler-wire pressure measurements, to implement the recommended current best practice methodologies in terms of patient-specific geometry and boundary conditions. The study aimed to evaluate and verify this workflow through detailed qualitative and quantitative comparisons of the CFD and in vivo data. Based on data acquired from five TBAD patients, a range of essential model inputs was obtained, including inlet flow waveforms and 3-element Windkessel model parameters, which can be utilised in further studies where in vivo flow data is not available. Local and global analysis showed good consistency between CFD results and 4D-MRI data, with the maximum velocity in the primary entry tear differing by up to 0.3 m/s, and 80% of the analysed regions achieving moderate or strong correlations between the predicted and in vivo velocities. CFD predicted pressures were generally well matched to the Doppler-wire measurements, with some deviation in peak systolic values. Overall, this study presents a validated comprehensive workflow with extensive data for CFD simulation of TBAD.

Epidemiological studies have demonstrated that ambient air pollution has been closely associated with cardiovascular diseases. Carbon monoxide (CO) is a common ambient air pollutant that can cause adverse effects on the heart. CO is known to cause tissue ischemia, resulting in ventricular arrhythmias. However, accumulating biological studies showed that CO could exert effects on multiple cardiac ionic channels under normoxic conditions, which might indicate new proarrhythmic mechanisms other than ischemia-mediated electrophysiology changes. In this work, we evaluated the functional impacts of CO on human ventricles using a multi-scale model of human ventricular tissue. Experimental data regarding the effects of CO on different ion channels were incorporated into the cell model to explore the alterations of ventricular electrophysiology. Simulation results suggested that CO significantly prolonged the duration of ventricular action potentials, enhanced the transmural dispersion of repolarization, and reduced the adaptability of ventricular tissue to fast heart rates. In addition, simulated pseudo-ECGs showed consistent manifestations with the clinical observation that CO caused an apparent QT interval prolongation and T-wave widening, indicating that CO affected the heart's abnormal ventricular repolarization.

BACKGROUND: We implemented a computerized decision support system (CDSS) integrated in the in-house computerized physician order entry (CPOE) system to assist physicians with antimicrobial prescribing decisions in the context of the multicenter cluster-randomized COMPASS trial (NCT03120975). Some physicians in the intervention wards complained about the perceived extra-time associated with the use of the CDSS compared with routine prescribing through CPOE. The aim of this study was to compare the time needed to prescribe antimicrobials with and without the CDSS. METHODS: Physicians with and without previous experience with the COMPASS CDSS working at our hospital in Geneva, Switzerland, were recruited to prescribe antimicrobials using clinical vignettes. Physicians without experience received a brief explanation of the CDSS. Each physician received 2 groups of 5-7 clinical vignettes randomly selected from a pool of 28. Each group of vignettes included prescriptions with different levels of complexity (empiric versus targeted or pre-defined treatment, dose adjustment for renal function, oral switch, treatments for which COMPASS does not provide recommendations or where a deviation was necessary). Prescriptions were completed using the standard CPOE (first set), then using COMPASS (second set). A print version of the local antimicrobial guidelines was available for consultation. Time to complete each prescription was recorded (including time needed to consult paper guidelines). The Mann-Whitney test was used for comparisons. Consultation of guidelines booklet and concordance with local guidelines were assessed. RESULTS: Twenty-five physicians were recruited. Thirteen (52%) had previously used COMPASS. Among them, 11 (85%) estimated the extra-time being above 1 min. We evaluated a total of 296 vignettes. Overall, the median time to complete a prescription was 55.5 s (IQR 38-86) using COMPASS and 50 s (IQR 31-88) using the standard CPOE (p = 0.24). Concordance of prescriptions with local guidelines was similar with the 2 systems (127/148, 85.8% for both), but consultation of paper guidelines was more frequent when prescribing without the CDSS (49.3% (73/148) vs 22.3% (33/148)). CONCLUSIONS: The increased time required for prescribing using COMPASS is overestimated by end-users. Information collected in the study will be used to streamline the prescribing process via COMPASS and increase acceptance.

Deep learning (DL) has shown great success in the field of medical image analysis. In the wake of the current pandemic situation of SARS-CoV-2, a few pioneering works based on DL have made significant progress in automated screening of COVID-19 disease from the chest X-ray (CXR) images. But these DL models have no inherent way of expressing uncertainty associated with the model's prediction, which is very important in medical image analysis. Therefore, in this paper, we develop an uncertainty-aware convolutional neural network model, named UA-ConvNet, for the automated detection of COVID-19 disease from CXR images, with an estimation of associated uncertainty in the model's predictions. The proposed approach utilizes the EfficientNet-B3 model and Monte Carlo (MC) dropout, where an EfficientNet-B3 model has been fine-tuned on the CXR images. During inference, MC dropout has been applied for M forward passes to obtain the posterior predictive distribution. After that mean and entropy have been calculated on the obtained predictive distribution to get the mean prediction and model uncertainty. The proposed method is evaluated on the three different datasets of chest X-ray images, namely the COVID19CXr, X-ray image, and Kaggle datasets. The proposed UA-ConvNet model achieves a G-mean of 98.02% (with a Confidence Interval (CI) of 97.99-98.07) and sensitivity of 98.15% for the multi-class classification task on the COVID19CXr dataset. For binary classification, the proposed model achieves a G-mean of 99.16% (with a CI of 98.81-99.19) and a sensitivity of 99.30% on the X-ray Image dataset. Our proposed approach shows its superiority over the existing methods for diagnosing the COVID-19 cases from the CXR images.

This systematic review provides researchers interested in feature selection (FS) for processing microarray data with comprehensive information about the main research directions for gene expression classification conducted during the recent seven years. A set of 132 researches published by three different publishers is reviewed. The studied papers are categorized into nine directions based on their objectives. The FS directions that received various levels of attention were then summarized. The review revealed that 'propose hybrid FS methods' represented the most interesting research direction with a percentage of 34.9%, while the other directions have lower percentages that ranged from 13.6% down to 3%. This guides researchers to select the most competitive research direction. Papers in each category are thoroughly reviewed based on six perspectives, mainly: method(s), classifier(s), dataset(s), dataset dimension(s) range, performance metric(s), and result(s) achieved.

Patients on hemodialysis (HD) are known to be at an increased risk of mortality. Hypoalbuminemia is one of the most important risk factors of death in HD patients, and is an independent risk factor for all-cause mortality that is associated with cardiac death, infection, and Protein-Energy Wasting (PEW). It is a clinical challenge to elevate serum albumin level. In addition, predicting trends in serum albumin level is effective for personalized treatment of hypoalbuminemia. In this study, we analyzed a total of 3069 records collected from 314 HD patients using a machine learning method that is based on an improved binary mutant quantum grey wolf optimizer (MQGWO) combined with Fuzzy K-Nearest Neighbor (FKNN). The performance of the proposed MQGWO method was evaluated using a series of experiments including global optimization experiments, feature selection experiments on open data sets, and prediction experiments on an HD dataset. The experimental results showed that the most critical relevant indicators such as age, presence or absence of diabetes, dialysis vintage, and baseline albumin can be identified by feature selection. Remarkably, the accuracy and the specificity of the method were 98.39% and 96.77%, respectively, demonstrating that this model has great potential to be used for detecting serum albumin level trends in HD patients.

As an emerging resource, Gram-negative Burkholderia bacteria were able to produce a wide range of bioactive secondary metabolites with potential therapeutic and biotechnological applications. Genome mining has emerged as an influential platform for screening and pinpointing natural product diversity with the increasing number of Burkholderia genome sequences. Here, for genome mining of potential biosynthetic gene clusters (BGCs) and prioritizing prolific producing Burkholderia strains, we investigated the relationship between species evolution and distribution of main BGC groups using computational analysis of complete genome sequences of 248 Burkholderia species publicly available. We uncovered significantly differential distribution patterns of BGCs in the Burkholderia phyla, even among strains that are genetically very similar. We found various types of BGCs in Burkholderia, including some representative and most common BGCs for biosynthesis of encrypted or known terpenes, non-ribosomal peptides (NRPs) and some hybrid BGCs for cryptic products. We also observed that Burkholderia contain a lot of unspecified BGCs, representing high potentials to produce novel compounds. Analysis of BGCs for RiPPs (Ribosomally synthesized and posttranslationally modified peptides) and a texobactin-like BGC as examples showed wide classification and diversity of RiPP BGCs in Burkholderia at species level and metabolite predication. In conclusion, as the biggest investigation in silico by far on BGCs of the particular genus Burkholderia, our data implied a great diversity of natural products in Burkholderia and BGC distributions closely related to phylogenetic variation, and suggested different or concurrent strategies used to identify new drug molecules from these microorganisms will be important for the selection of potential BGCs and prolific producing strains for drug discovery.