COVID-19 lung infection segmentation from chest CT images based on CAPA-ResUNet

Coronavirus disease 2019 (COVID-19) epidemic has devastating effects on personal health around the world. It is significant to achieve accurate segmentation of pulmonary infection regions, which is an early indicator of disease. To solve this problem, a deep learning model, namely, the content-aware pre-activated residual UNet (CAPA-ResUNet), was proposed for segmenting COVID-19 lesions from CT slices. In this network, the pre-activated residual block was used for down-sampling to solve the problems of complex foreground and large fluctuations of distribution in datasets during training and to avoid gradient disappearance. The area loss function based on the false segmentation regions was proposed to solve the problem of fuzzy boundary of the lesion area. This model was evaluated by the public dataset (COVID-19 Lung CT Lesion Segmentation Challenge—2020) and compared its performance with those of classical models. Our method gains an advantage over other models in multiple metrics. Such as the Dice coefficient, specificity (Spe), and intersection over union (IoU), our CAPA-ResUNet obtained 0.775 points, 0.972 points, and 0.646 points, respectively. The Dice coefficient of our model was 2.51% higher than Content-aware residual UNet (CARes-UNet). The code is available at https://github.com/malu108/LungInfectionSeg. © 2022 Wiley Periodicals LLC.

One shot learning approach for cross spectrum periocular verification

The use of face mask during the COVID-19 pandemic has increased the popularity of the periocular biometrics in surveillance applications. Despite of the rapid advancements in this area, matching images over cross spectrum is still a challenging problem. Reason may be two-fold 1) variations in image illumination 2) small size of available data sets and/or class imbalance problem. This paper proposes Siamese architecture based convolutional neural networks which works on the concept of one-shot classification. In one shot classification, network requires a single training example from each class to train the complete model which may lead to reduce the need of large dataset as well as doesn't matter whether the dataset is imbalance. The proposed architectures comprise of identical subnetworks with shared weights whose performance is assessed on three publicly available databases namely IMP, UTIRIS and PolyU with four different loss functions namely Binary cross entropy loss, Hinge loss, contrastive loss and Triplet loss. In order to mitigate the inherent illumination variations of cross spectrum images CLAHE was used to preprocess images. Extensive experiments show that the proposed Siamese CNN model with triplet loss function outperforms the states of the art periocular verification methods for cross, mono and multi spectral periocular image matching. © 2023, The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.

Masked face recognition with convolutional visual self-attention network

With the global outbreak of COVID-19, wearing face masks has been actively introduced as an effective public measure to reduce the risk of virus infection. This measure leads to the failure of face recognition in many cases. Therefore, it is very necessary to improve the recognition performance of masked face recognition (MFR). Inspired by the successful application of self-attention in computer vision, we propose a Convolutional Visual Self-Attention Network (CVSAN), which uses self-attention to augment the convolution operator. Specifically, this is achieved by connecting a convolutional feature map, which enforces local features, to a self-attention feature map that is capable of modeling long-range dependencies. Since there is currently no publicly available large-scale masked face data, we generate a Masked VGGFace2 dataset based on the face detection algorithm to train the CVSAN model. Experiments show that the CVSAN algorithm significantly improves the performance of MFR compared to other algorithms. © 2022

CoVSeverity-Net: an efficient deep learning model for COVID-19 severity estimation from Chest X-Ray images

Purpose: COVID-19 is not going anywhere and is slowly becoming a part of our life. The World Health Organization declared it a pandemic in 2020, and it has affected all of us in many ways. Several deep learning techniques have been developed to detect COVID-19 from Chest X-Ray images. COVID-19 infection severity scoring can aid in establishing the optimum course of treatment and care for a positive patient, as all COVID-19 positive patients do not require special medical attention. Still, very few works are reported to estimate the severity of the disease from the Chest X-Ray images. The unavailability of the large-scale dataset might be a reason. Methods: We aim to propose CoVSeverity-Net, a deep learning-based architecture for predicting the severity of COVID-19 from Chest X-ray images. CoVSeverity-Net is trained on a public COVID-19 dataset, curated by experienced radiologists for severity estimation. For that, a large publicly available dataset is collected and divided into three levels of severity, namely Mild, Moderate, and Severe. Results: An accuracy of 85.71% is reported. Conducting 5-fold cross-validation, we have obtained an accuracy of 87.82 ± 6.25%. Similarly, conducting 10-fold cross-validation we obtained accuracy of 91.26 ± 3.42. The results were better when compared with other state-of-the-art architectures. Conclusion: We strongly believe that this study has a high chance of reducing the workload of overworked front-line radiologists, speeding up patient diagnosis and treatment, and easing pandemic control. Future work would be to train a novel deep learning-based architecture on a larger dataset for severity estimation. © 2023, The Author(s), under exclusive licence to The Brazilian Society of Biomedical Engineering.

Camera-Based Blood Pressure Estimation via Windkessel Model and Waveform Features

Blood pressure (BP) is generally regarded as the vital sign most strongly correlated with human health. However, for decades, BP measurement has involved a cuff, which causes discomfort and even carries a risk of infection, given the current prevalence of COVID-19. Some studies address these problems using remote photoplethysmography (rPPG), which has shown great success in heart rate detection. Nevertheless, these approaches are not robust, and few have been evaluated with a sufficiently large dataset. We propose an rPPG-based BP estimation algorithm that predicts BP by leveraging the Windkessel model and hand-crafted waveform characteristics. A waveform processing procedure is presented for the rPPG signals to obtain a robust waveform template and thus extract BP-related features. Redundant and unstable features are eliminated via Monte Carlo simulation and according to their relationship with latent parameters (LSs) in the Windkessel model. For a comprehensive evaluation, the Chiao Tung BP (CTBP) dataset was constructed. The experiment was conducted over a four-week period of time to evaluate the validity period of the personalization in our system. On all the data, the proposed method outperforms the benchmark algorithms and yields mean absolute errors (MAEs) of 6.48 and 5.06 mmHg for systolic BP (SBP) and diastolic BP (DBP), respectively. The performance achieves a 'B' grade according to the validation protocol from the British Hypertension Society (BHS) for both SBP and DBP. © 1963-2012 IEEE.

A Hybrid Deep Fused Learning Approach to Segregate Infectious Diseases

Humankind is facing another deadliest pandemic of all times in history, caused by COVID-19. Apart from this challenging pandemic, World Health Organization (WHO) considers tuberculosis (TB) as a preeminent infectious disease due to its high infection rate. Generally, both TB and COVID-19 severely affect the lungs, thus hardening the job of medical practitioners who can often misidentify these diseases in the current situation. Therefore, the time of need calls for an immediate and meticulous automatic diagnostic tool that can accurately discriminate both diseases. As one of the preliminary smart health systems that examine three clinical states (COVID-19, TB, and normal cases), this study proposes an amalgam of image filtering, data-augmentation technique, transfer learning-based approach, and advanced deep-learning classifiers to effectively segregate these diseases. It first employed a generative adversarial network (GAN) and Crimmins speckle removal filter on X-ray images to overcome the issue of limited data and noise. Each pre-processed image is then converted into red, green, and blue (RGB) and Commission Internationale de l'Elcairage (CIE) color spaces from which deep fused features are formed by extracting relevant features using DenseNet121 and ResNet50. Each feature extractor extracts 1000 most useful features which are then fused and finally fed to two variants of recurrent neural network (RNN) classifiers for precise discrimination of three-clinical states. Comparative analysis showed that the proposed Bi-directional long-short-term-memory (Bi-LSTM) model dominated the long-short-term-memory (LSTM) network by attaining an overall accuracy of 98.22% for the three-class classification task, whereas LSTM hardly achieved 94.22% accuracy on the test dataset. © 2023 Tech Science Press. All rights reserved.

Perspectives of non-expert users on cyber security and privacy: An analysis of online discussions on twitter

Many researchers have studied non-expert users' perspectives of cyber security and privacy aspects of computing devices at home, but their studies are mostly small-scale empirical studies based on online surveys and interviews and limited to one or a few specific types of devices, such as smart speakers. This paper reports our work on an online social media analysis of a large-scale Twitter dataset, covering cyber security and privacy aspects of many different types of computing devices discussed by non-expert users in the real world. We developed two new machine learning based classifiers to automatically create the Twitter dataset with 435,207 tweets posted by 337,604 non-expert users in January and February of 2019, 2020 and 2021. We analyzed the dataset using both quantitative (topic modeling and sentiment analysis) and qualitative analysis methods, leading to various previously unknown findings. For instance, we observed a sharp (more than doubled) increase of non-expert users' tweets on cyber security and privacy during the pandemic in 2021, compare to in the pre-COVID years (2019 and 2020). Our analysis revealed a diverse range of topics discussed by non-expert users, including VPNs, Wi-Fi, smartphones, laptops, smart home devices, financial security, help-seeking, and roles of different stakeholders. Overall negative sentiment was observed across almost all topics in all the three years. Our results indicate the multi-faceted nature of non-expert users' perspectives on cyber security and privacy and call for more holistic, comprehensive and nuanced research on their perspectives. © 2022

Enhanced framework for COVID-19 prediction with computed tomography scan images using dense convolutional neural network and novel loss function

Recent studies have shown that computed tomography (CT) scan images can characterize COVID-19 disease in patients. Several deep learning (DL) methods have been proposed for diagnosis in the literature, including convolutional neural networks (CNN). But, with inefficient patient classification models, the number of ‘False Negatives' can put lives at risk. The primary objective is to improve the model so that it does not reveal ‘Covid' as ‘Non-Covid'. This study uses Dense-CNN to categorize patients efficiently. A novel loss function based on cross-entropy has also been used to improve the CNN algorithm's convergence. The proposed model is built and tested on a recently published large dataset. Extensive study and comparison with well-known models reveal the effectiveness of the proposed method over known methods. The proposed model achieved a prediction accuracy of 93.78%, while false-negative is only 6.5%. This approach's significant advantage is accelerating the diagnosis and treatment of COVID-19. © 2022

Curriculum Contrastive Learning for COVID-19 FAQ Retrieval

Medical Frequently Asked Question (FAQ) retrieval aims to find the most relevant question-answer pairs for a given user query, which is of great significance for enhancing people medical health awareness and knowledge. However, due to medical data privacy and labor-intensive labeling, there is a lack of large-scale question-matching training datasets. Previous methods directly use the collected question-answer pairs on search engines to train retrieval models, which achieved poor performance. Inspired by recent advances in contrastive learning, we propose a novel contrastive curriculum learning framework for modeling user medical queries. First, we design different data augmentation methods to generate positive samples and different types of negative samples. Second, we propose a curriculum learning strategy that associates difficulty levels with negative samples. Through a contrastive learning process from easy to hard, our method achieves excellent results on two medical datasets. © 2022 IEEE.

VQC-COVID-NET: Vector Quantization Contrastive Learning for Covid-19 Image Base Classification

Today, the COVID-19 epidemic has become extremely widespread. The first step in combating COVID-19 is identifying cases of infection. Real-time reverse transcriptase polymerase chain reaction is the most common method for identifying COVID (RT-PCR). This method, however, has been compromised by a time-consuming, laborious, and complex manual process. In addition to the RT-PCR test, screening computed tomography scan (CT) or X-ray images may be used to identify positive COVID-19 results, which could aid in the detection of COVID-19. Because of the continuing increase in new infections, the development of automated techniques for COVID-19 detection utilizing CT images is in high demand. This will aid in clinical diagnosis and alleviate the arduous task of image interpretation. Aggregating instances from various medical systems is highly advantageous for enlarging datasets for the development of machine learning techniques and the acquisition of robust, generalizable models. This study proposes a novel method for addressing distinct feature normalization in latent space due to cross-site domain shift in order to accurately execute COVID-19 identification using heterogeneous datasets with distribution disagreement. We propose using vector quantization to enhance the domain invariance of semantic embeddings in order to enhance classification performance on each dataset. We use two large, publicly accessible COVID-19 diagnostic CT scan datasets to develop and validate our proposed model. The experimental results demonstrate that our proposed method routinely outperforms state-of-the-art techniques on testing datasets. Public access to the implementation of our proposed method is available at https://github.com/khaclinh/VQC-COVID-NET. © 2022 IEEE.

Feasibility Study on Browser-Based Federated Machine Learning (FML) Architecture for Medical Application

Traditional approaches to Artificial Intelligence (AI) based medical image classification requires huge amounts of data sets to be stored in a centralized server for analysis and training. In medical applications, data privacy and ownership may pose a challenge. In addition, costs incurred by data transfer and cloud server may pose a challenge to implementing a large dataset. This work studies the feasibility of a decentralized, browser-based Artificial Intelligence (AI) federated machine learning (FML) architecture. The proposed work studies the feasibility of bringing training and inference to the browser, hence removing the need to transfer raw data to a centralized server. If feasible, the system allows practitioners to compress and upload their pre-trained model to the server instead of raw data. This allows medical practitioners to update the model without the need to reveal their raw data. A sandbox system was implemented by applying transfer learning on MobileNet V3 and was tested with chest X-ray image datasets from COVID-19, viral pneumonia, and normal patients to simulate medical usage environment. The training speed, model performance and inference speed were tested on a PC browser and mobile phone with various levels of network throttling and image degradation. © 2022 IEEE.

YOLO Based Thermal Screening Using Artificial Intelligence (Al) for Instinctive Human Facial Detection

Covid19 remains the world's greatest public health emergency. It has become indispensable to measure the temperature of people entering or leaving croweded places to ease the identification of potentially infected and to isolate them from spreading and preventing the spread of the ongoing global pandemic of coronavirus disease. This research work is focusing on thermal screening for an automated scanner using Artificial Intelligence (AI) for instinctive temperature measurement on human faces. The framework used for facial detection is known as YOLOv5 which is a family of compound-scaled object detection models trained on the COCO, a large-scale object detection, segmentation, and captioning dataset. YOLOv5 is able to detect several different objects simultaneously by using its available pre-trained models and robustness of detecting faces even at the vicinity of face masks. The research presents the application, training procedure and capability of the Yolov5. This system is not only used for the human face detection, but also for the detection of some commonly-used objects as an extension to its overall application and performance. Yolov5 is readily available to be implemented in Python, the core programming language working under an Ubuntu-based Operating System providing users the best experience. One of the important outcomes of this research work is the development of a Graphical User Interface (GUI) to work alongside the main programme flow. © 2022 IEEE.

Mask usage recognition using vision transformer with transfer learning and data augmentation

The COVID-19 pandemic has disrupted various levels of society. The use of masks is essential in preventing the spread of COVID-19 by identifying an image of a person using a mask. Although only 23.1% of people use masks correctly, Artificial Neural Networks (ANN) can help classify the use of good masks to help slow the spread of the Covid-19 virus. However, it requires a large dataset to train an ANN that can classify the use of masks correctly. MaskedFace-Net is a suitable dataset consisting of 137016 digital images with 4 class labels, namely Mask, Mask Chin, Mask Mouth Chin, and Mask Nose Mouth. Mask classification training utilizes Vision Transformers (ViT) architecture with transfer learning method using pre-trained weights on ImageNet-21k, with random augmentation. In addition, the hyper-parameters of training of 20 epochs, an Stochastic Gradient Descent (SGD) optimizer with a learning rate of 0.03, a batch size of 64, a Gaussian Cumulative Distribution (GeLU) activation function, and a Cross-Entropy loss function are used to be applied on the training of three architectures of ViT, namely Base-16, Large-16, and Huge-14. Furthermore, comparisons of with and without augmentation and transfer learning are conducted. This study found that the best classification is transfer learning and augmentation using ViT Huge-14. Using this method on MaskedFace-Net dataset, the research reaches an accuracy of 0.9601 on training data, 0.9412 on validation data, and 0.9534 on test data. This research shows that training the ViT model with data augmentation and transfer learning improves classification of the mask usage, even better than convolutional-based Residual Network (ResNet). © 2023 The Author(s)

RoCoFake – A Romanian Covid-19 Fake News Dataset

The Covid-19 pandemic, managed to shed light onto a neglected problem – that of fake news. Even though lockdowns were imposed in most parts of the world, collaboration between researchers across the globe wasn't impeded. Moreover, the lockdown has deprived people of face-to-face interactions and so they shifted towards online communication. This translated into a massive chatting data, which part was true, but fake information also had its share. Therefore, it is of great interest to develop a dataset to try to spot the fake information. RoCoFake comes to address the lack of resources in this domain, by providing a Romanian Covid-19 Fake News dataset, by aggregating various resources available online, like tweets, news titles and fact-checking news sites like factual.ro. This data provides researchers from the medical domain particularly, but not only, with a valuable, open-access data source useful for various research projects. A benchmark for fake news detection is also provided, so that future investigations can compare against our research. Results suggest that even though the dataset is relatively large, improvements can be made by incorporating retweets and comments. © 2022 IEEE.

Architecture-based Evaluation of VGG16 and ResNet Models for an Online Deep Learning Environment for Medical Applications

Image classification using deep learning models has evolved impressively well in the past decade. Datasets containing millions of images grouped into thousands of classes have been used to train and test these models. Medical image classification however still faces the challenge of scarcity in datasets. Gathering data from various locations and placing it in a commonly accessed dataset is highly time-consuming. Diseases need real-Time response just like any other mission-critical operation and online deep learning could be handy. There are many pre-Trained models which acquired good accuracy on large datasets. But as the depth of the model increases the time taken to train the model and the number of computations also increase. In this paper, we evaluated two models with different architectures. VGG16 is a 16-layer normal stack of convolutional layers and ResNet50V2 is a stack of residual blocks with skip connections and 50 layers. We used a Computer Tomography (CT) Lung image dataset to classify images into COVID, healthy and pneumonia images. We found that VGG16 is taking lesser time and computations with reduced loss when compared to the ResNet50V2 model. We finally conclude that ResNet50V2 is taking more time to train images as the model is 50 layers deep, whereas the VGG16 model is only 16 layers deep. Also, images that show mild infection were predicted as healthy images by ResNet50V2 but predicted correctly by the VGG16 model. © 2022 IEEE.

Real-Time Mask Detection Based on SSD-MobileNetV2

After the outbreak of COVID-19, mask detection, as the most convenient and effective means of prevention, plays a crucial role in epidemic prevention and control. An excellent automatic real-time mask detection system can reduce a lot of work pressure for relevant staff. However, by analyzing the existing mask detection approaches, we find that they are mostly resource-intensive and do not achieve a good balance between speed and accuracy. And there is no perfect face mask dataset at present. In this paper, we propose a new architecture for mask detection. Our system uses SSD as the mask locator and classifier, and further replaces VGG-16 with MobileNetV2 to extract the features of the image and reduce a lot of parameters. Therefore, our system can be deployed on embedded devices. Transfer learning methods are used to transfer pre-trained models from other domains to our model. Data enhancement methods in our system such as MixUp effectively prevent overfitting. It also effectively reduces the dependence on large-scale datasets. By doing experiments in practical scenarios, the results demonstrate that our system performed well in real-time mask detection. © 2022 IEEE.