RESUMO
We present a synthetic augmentation approach towards improving monocular face presentation-attack-detection (PAD) robustness to real-world noise additions. Face PAD algorithms secure authentication systems against spoofing attacks, such as pictures, videos, and 2D-inspired masks. Best-in-class PAD methods typically use 3D imagery, but these can be expensive. To reduce application cost, there is a growing field investigating monocular algorithms that detect facial artifacts. These approaches work well in laboratory conditions, but can be sensitive to the imaging environment (e.g., sensor noise, dynamic lighting, etc.). The ideal solution for noise robustness is training under all expected conditions; however, this is time consuming and expensive. Instead, we propose that physics-informed noise-augmentations can pragmatically achieve robustness. Our toolbox contains twelve sensor and lighting effect generators. We demonstrate that our toolbox generates more robust PAD features than popular augmentation methods in noisy test-evaluations. We also observe that the toolbox improves accuracy on clean test data, suggesting that it inherently helps discern spoof artifacts from imaging artifacts. We validate this hypothesis through an ablation study, where we remove liveliness pairs (e.g., live or spoof imagery only for participants) to identify how much real data can be replaced with synthetic augmentations. We demonstrate that using these noise augmentations allows us to achieve better test accuracy while only requiring 30% of participants to be fully imaged under all conditions. These findings indicate that synthetic noise augmentations are a great way to improve PAD, addressing noise robustness while simplifying data collection.
RESUMO
HIV infects the CD4 cells which marks the suppression of our immune system. DNA from serum of healthy, treated and untreated HIV infected individuals was extracted. The DNA was subjected to 16S metagenomic sequencing and analyzed using QIIME2 pipeline. 16S sequencing analysis showed serum microbiome was dominated by Firmicutes, Proteobacteria, Bacteroidota and Actinobacteria. Treated HIV infection showed highest abundance of Firmicutes (66.40%) significantly higher than untreated HIV infection (35.88%) and control (41.89%). Bacilli was most abundant class in treated (63.59%) and second most abundant in untreated (34.53%) while control group showed highest abundance of class Gamma-proteobacteria (45.86%). Untreated HIV infection group showed Enterococcus (10.72%) and Streptococcus (6.599%) as the most abundant species. Untreated HIV infection showed significantly higher (p = 0.0039) species richness than treated and control groups. An altered serum microbiome of treated HIV infection and higher microbial abundance in serum of untreated HIV infection was observed.
Assuntos
Infecções por HIV , Microbiota , Infecções por HIV/genética , Humanos , Metagenoma , Metagenômica , RNA Ribossômico 16S/genéticaRESUMO
The use of face masks has increased dramatically since the COVID-19 pandemic started in order to to curb the spread of the disease. Additionally, breakthrough infections caused by the Delta and Omicron variants have further increased the importance of wearing a face mask, even for vaccinated individuals. However, the use of face masks also induces attenuation in speech signals, and this change may impact speech processing technologies, e.g., automated speaker verification (ASV) and speech to text conversion. In this paper we examine Automatic Speaker Verification (ASV) systems against the speech samples in the presence of three different types of face mask: surgical, cloth, and filtered N95, and analyze the impact on acoustics and other factors. In addition, we explore the effect of different microphones, and distance from the microphone, and the impact of face masks when speakers use ASV systems in real-world scenarios. Our analysis shows a significant deterioration in performance when an ASV system encounters different face masks, microphones, and variable distance between the subject and microphone. To address this problem, this paper proposes a novel framework to overcome performance degradation in these scenarios by realigning the ASV system. The novelty of the proposed ASV framework is as follows: first, we propose a fused feature descriptor by concatenating the novel Ternary Deviated overlapping Patterns (TDoP), Mel Frequency Cepstral Coefficients (MFCC), and Gammatone Cepstral Coefficients (GTCC), which are used by both the ensemble learning-based ASV and anomaly detection system in the proposed ASV architecture. Second, this paper proposes an anomaly detection model for identifying vocal samples produced in the presence of face masks. Next, it presents a Peak Norm (PN) filter to approximate the signal of the speaker without a face mask in order to boost the accuracy of ASV systems. Finally, the features of filtered samples utilizing the PN filter and samples without face masks are passed to the proposed ASV to test for improved accuracy. The proposed ASV system achieved an accuracy of 0.99 and 0.92, respectively, on samples recorded without a face mask and with different face masks. Although the use of face masks affects the ASV system, the PN filtering solution overcomes this deficiency up to 4%. Similarly, when exposed to different microphones and distances, the PN approach enhanced system accuracy by up to 7% and 9%, respectively. The results demonstrate the effectiveness of the presented framework against an in-house prepared, diverse Multi Speaker Face Masks (MSFM) dataset, (IRB No. FY2021-83), consisting of samples of subjects taken with a variety of face masks and microphones, and from different distances.
Assuntos
COVID-19 , Humanos , Máscaras , Pandemias/prevenção & controle , SARS-CoV-2 , FalaRESUMO
The modern-day vehicle is evolved in a cyber-physical system with internal networks (controller area network (CAN), Ethernet, etc.) connecting hundreds of micro-controllers. From the traditional core vehicle functions, such as vehicle controls, infotainment, and power-train management, to the latest developments, such as advanced driver assistance systems (ADAS) and automated driving features, each one of them uses CAN as their communication network backbone. Automated driving and ADAS features rely on data transferred over the CAN network from multiple sensors mounted on the vehicle. Verifying the integrity of the sensor data is essential for the safety and security of occupants and the proper functionality of these applications. Though the CAN interface ensures reliable data transfer, it lacks basic security features, including message authentication, which makes it vulnerable to a wide array of attacks, including spoofing, replay, DoS, etc. Using traditional cryptography-based methods to verify the integrity of data transmitted over CAN interfaces is expected to increase the computational complexity, latency, and overall cost of the system. In this paper, we propose a light-weight alternative to verify the sensor data's integrity for vehicle applications that use CAN networks for data transfers. To this end, a framework for 2-dimensional quantization index modulation (2D QIM)-based data hiding is proposed to achieve this goal. Using a typical radar sensor data transmission scenario in an autonomous vehicle application, we analyzed the performance of the proposed framework regarding detecting and localizing the sensor data tampering. The effects of embedding-induced distortion on the applications using the radar data were studied through a sensor fusion algorithm. It was observed that the proposed framework offers the much-needed data integrity verification without compromising on the quality of sensor fusion data and is implemented with low overall design complexity. This proposed framework can also be used on any physical network interface other than CAN, and it offers traceability to in-vehicle data beyond the scope of the in-vehicle applications.
RESUMO
Naegleria fowleri, also known as brain-earing amoeba, causes severe and rapidly fatal CNS infection in humans called primary amebic meningoencephalitis (PAM). The DNA from the N. fowleri clinical isolate was sequenced for circular extrachromosomal ribosomal DNA (CERE - rDNA). The CERE contains 18 S, 5.8 S, and 28 S ribosomal subunits separated by internal transcribed spacers, 5 open reading frames (ORFs), and mostly repeat elements comprising 7268 bp out of 15,786 bp (46%). A wide variety of variations and recombination events were observed. Finally, the ORFs that comprised only 4 hypothetical proteins were modeled and screened against Zinc drug-like compounds. Two compounds [ZINC77564275 (ethyl 2-(((4-isopropyl-4 H-1,2,4-triazol-3-yl) methyl) (methyl)amino) oxazole-4-carboxylate) and ZINC15022129 (5-(2-methoxyphenoxy)-[2,2'-bipyrimidine]-4,6(1 H,5 H)-dione)] were finalized as potential druggable compounds based on ADME toxicity analysis. We propose that the compounds showing the least toxicity would be potential drug candidates after laboratory experimental validation is performed.
Assuntos
DNA Ribossômico , Sequenciamento de Nucleotídeos em Larga Escala , Naegleria fowleri , Naegleria fowleri/genética , Humanos , DNA Ribossômico/genética , Encéfalo/metabolismo , Genótipo , Fases de Leitura AbertaRESUMO
Drug-to-drug interaction (DDIs) occurs when a patient consumes multiple drugs. Therefore, it is possible that any medication can influence other drugs' effectiveness. The drug-to-drug interactions are detected based on the interactions of chemical substructures, targets, pathways, and enzymes; therefore, machine learning (ML) and deep learning (DL) techniques are used to find the associated DDI events. The DL model, i.e., Convolutional Neural Network (CNN), is used to analyze the DDI. DDI is based on the 65 different drug-associated events, which is present in the drug bank database. Our model uses the inputs, which are chemical structures (i.e., smiles of drugs), enzymes, pathways, and the target of the drug. Therefore, for the multi-model CNN, we use several layers, activation functions, and features of drugs to achieve better accuracy as compared to traditional prediction algorithms. We perform different experiments on various hyperparameters. We have also carried out experiments on various iterations of drug features in different sets. Our Multi-Modal Convolutional Neural Network - Drug to Drug Interaction (MCNN-DDI) model achieved an accuracy of 90.00% and an AUPR of 94.78%. The results showed that a combination of the drug's features (i.e., chemical substructure, target, and enzyme) performs better in DDIs-associated events prediction than other features.
Assuntos
Algoritmos , Redes Neurais de Computação , Humanos , Interações Medicamentosas , Aprendizado de MáquinaRESUMO
In recent times, deep learning has emerged as a great resource to help research in medical sciences. A lot of work has been done with the help of computer science to expose and predict different diseases in human beings. This research uses the Deep Learning algorithm Convolutional Neural Network (CNN) to detect a Lung Nodule, which can be cancerous, from different CT Scan images given to the model. For this work, an Ensemble approach has been developed to address the issue of Lung Nodule Detection. Instead of using only one Deep Learning model, we combined the performance of two or more CNNs so they could perform and predict the outcome with more accuracy. The LUNA 16 Grand challenge dataset has been utilized, which is available online on their website. The dataset consists of a CT scan with annotations that better understand the data and information about each CT scan. Deep Learning works the same way our brain neurons work; therefore, deep learning is based on Artificial Neural Networks. An extensive CT scan dataset is collected to train the deep learning model. CNNs are prepared using the data set to classify cancerous and non-cancerous images. A set of training, validation, and testing datasets is developed, which is used by our Deep Ensemble 2D CNN. Deep Ensemble 2D CNN consists of three different CNNs with different layers, kernels, and pooling techniques. Our Deep Ensemble 2D CNN gave us a great result with 95% combined accuracy, which is higher than the baseline method.
Assuntos
Aprendizado Profundo , Neoplasias Pulmonares , Lesões Pré-Cancerosas , Humanos , Redes Neurais de Computação , Tomografia Computadorizada por Raios X/métodos , AlgoritmosRESUMO
Breast adenocarcinoma is the most common of all cancers that occur in women. According to the United States of America survey, more than 282,000 breast cancer patients are registered each year; most of them are women. Detection of cancer at its early stage saves many lives. Each cell contains the genetic code in the form of gene sequences. Changes in the gene sequences may lead to cancer. Replication and/or recombination in the gene base sometimes lead to a permanent change in the nucleotide sequence of the genome, called a mutation. Cancer driver mutations can lead to cancer. The proposed study develops a framework for the early detection of breast adenocarcinoma using machine learning techniques. Every gene has a specific sequence of nucleotides. A total of 99 genes are identified in various studies whose mutations can lead to breast adenocarcinoma. This study uses the dataset taken from 4127 human samples, including men and women from more than 12 cohorts. A total of 6170 mutations in gene sequences are used in this study. Decision Tree, Random Forest, and Gaussian Naïve Bayes are applied to these gene sequences using three evaluation methods: independent set testing, self-consistency testing, and tenfold cross-validation testing. Evaluation metrics such as accuracy, specificity, sensitivity, and Mathew's correlation coefficient are calculated. The decision tree algorithm obtains the best accuracy of 99% for each evaluation method.
Assuntos
Adenocarcinoma , Neoplasias da Mama , Adenocarcinoma/diagnóstico , Adenocarcinoma/genética , Teorema de Bayes , Neoplasias da Mama/diagnóstico , Neoplasias da Mama/genética , Carcinogênese , Carcinógenos , Feminino , Humanos , Aprendizado de Máquina , Masculino , MutaçãoRESUMO
In the current study, we have systematically analysed the mitochondrial DNA (mtDNA) sequence of Naegleria fowleri (N. fowleri) isolate AY27, isolated from Karachi, Pakistan. The N. fowleri isolate AY27 has a circular mtDNA (49,541 bp), which harbours 69 genes (46 protein-coding genes, 21 tRNAs and 2 rRNAs). The pan-genome analysis of N. fowleri species showed a Bpan value of 0.137048, which implies that the pan-genome is open. KEGG classified core, accessory and unique gene clusters for human disease, metabolism, environmental information processing, genetic information processing and organismal system. Similarly, COG characterization of protein showed that core and accessory genes are involved in metabolism, information storages and processing, and cellular processes and signaling. The Naegleria species (n = 6) formed a total of 47 gene clusters; 42 single-copy gene clusters and 5 orthologous gene clusters. It was noted that 100% genes of Naegleria species were present in the orthogroups. We identified 44 single nucleotide polymorphisms (SNP) in the N. fowleri isolate AY27 mtDNA using N. fowleri strain V511 as a reference. Whole mtDNA phylogenetic tree analysis showed that N. fowleri isolates AY27 is closely related to N. fowleri (Accession no. JX174181.1). The ANI (Average Nucleotide Identity) values presented a much clear grouping of the Naegleria species compared to the whole mtDNA based phylogenetic analysis. The current study gives a comprehensive understanding of mtDNA architecture as well as a comparison of Naegleria species (N. fowleri and N. gruberi species) at the mitochondrial genome sequence level.
Assuntos
Genoma Mitocondrial , Naegleria fowleri , Naegleria , DNA Mitocondrial/genética , DNA Mitocondrial/metabolismo , DNA de Protozoário , Evolução Molecular , Genoma Mitocondrial/genética , Naegleria/genética , Naegleria fowleri/genética , FilogeniaRESUMO
Single Nucleotide Polymorphisms (SNPs) are the most common candidate mutations in human beings that play a vital role in the genetic basis of certain diseases. Previous studies revealed that Solute Carrier Family 26 Member 4 (SLC26A4) being an essential gene of the multi-faceted transporter family SLC26 facilitates reflexive movement of Iodide into follicular lumen through apical membrane of thyrocyte. SLC26A4 gene encodes Pendred protein, a membrane glycoprotein, highly hydrophobic in nature, present at the apical membrane of thyrocyte functioning as transporter of iodide for thyroid cells. A minor genetic variation in SLC26A4 can cause Pendred syndrome, a syndrome associated with thyroid glands and deafness. In this study, we performed in-silico analysis of 674 missense SNPs of SLC26A4 using different computational platforms. The bunch of tools including SNPNEXUS, SNAP-2, PhD-SNP, SNPs&GO, I-Mutant, ConSurf, and ModPred were used to predict 23 highly confident damaging and disease causing nsSNPs (G209V, G197R, L458P, S427P, Q101P, W472R, N392Y, V359E, R409C, Q235R, R409P, G139V, G497S, H723R, D87G, Y127H, F667C, G334A, G95R, S427C, R291W, Q383H and E384G) that could potentially alter the SLC26A4 gene. Moreover, protein structure prediction, protein-ligand docking and Molecular Dynamics simulation were performed to confirm the impact of two evident alterations (Y127H and G334A) on the protein structure and function.
Assuntos
Biologia Computacional , Predisposição Genética para Doença , Polimorfismo de Nucleotídeo Único/genética , Transportadores de Sulfato/genética , Surdez/genética , Surdez/patologia , Bócio Nodular/genética , Bócio Nodular/patologia , Perda Auditiva Neurossensorial/genética , Perda Auditiva Neurossensorial/patologia , Humanos , Ligantes , Simulação de Dinâmica Molecular , Mutação/genética , Ligação Proteica , Conformação Proteica , Relação Estrutura-Atividade , Transportadores de Sulfato/químicaRESUMO
OBJECTIVES: Dengue epidemic is a dynamic and complex phenomenon that has gained considerable attention due to its injurious effects. The focus of this study is to statically analyze the nature of the dengue epidemic network in terms of whether it follows the features of a scale-free network or a random network. METHODS: A multifarious network of Aedes aegypti is addressed keeping the viewpoint of a complex system and modelled as a network. The dengue network has been transformed into a one-mode network from a two-mode network by utilizing projection methods. Furthermore, three network features have been analyzed, the power-law, clustering coefficient, and network visualization. In addition, five methods have been applied to calculate the global clustering coefficient. RESULTS: It has been observed that dengue epidemic follows a power-law, with the value of its exponent γ = -2.1. The value of the clustering coefficient is high for dengue cases, as weight of links. The minimum method showed the highest value among the methods used to calculate the coefficient. Network visualization showed the main areas. Moreover, the dengue situation did not remain the same throughout the observed period. CONCLUSIONS: The results showed that the network topology exhibits the features of a scale-free network instead of a random network. Focal hubs are highlighted and the critical period is found. Outcomes are important for the researchers, health officials, and policy makers who deal with arbovirus epidemic diseases. Zika virus and Chikungunya virus can also be modelled and analyzed in this manner.