Segmentation of COVID-19 Chest CT Images Based on SwishUnet

Coronavirus disease 2019 (COVID-19), the disease caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), has been spreading since 2019 until now. Chest CT-scan images have contributed significantly to the prognosis, diagnosis, and detection of complications in COVID-19. Automatic segmentation of COVID-19 infections involving ground-glass opacities and consolidation can assist radiologists in COVID-19 screening, which helps reduce time spent analyzing the infection. In this study, we proposed a novel deep learning network to segment lung damage caused by COVID-19 by utilizing EfficientNet and Resnet as the encoder and a modified U-Net with Swish activation, namely swishUnet, as the decoder. In particular, swishUnet allows the model to deal with smoothness, non-monotonicity, and one-sided boundedness at zero. Three experiments were conducted to evaluate the performance of the proposed architecture on the 100 CT scans and 9 volume CT scans from Italian the society of medical and interventional radiology. The results of the first experiment showed that the best sensitivity was 82.7% using the Resnet+swishUnet method with the Tversky loss function. In the second experiment, the architecture with basic Unet only got a sensitivity of 67.2. But with our proposed method, we can improve to 88.1% by using EfficientNet+SwishUnet. For the third experiment, the best performance sensitivity is Resnet+swishUnet with 79.8%. All models with SwishUnet have the same specificity where the value is 99.8%. From the experiments we conclude that our proposed method with SwishUnet encoder has better performance than the previous method © 2023, International Journal of Intelligent Engineering and Systems.All Rights Reserved.

Masked face recognition on mobile devices using deep learning

Deep Convolution Neural Network (DCNN) based facial recognition has made significant progress in recent years. Currently, facial recognition technology has emerged as an important authentication tool on mobile devices. Hence, a fast and lightweight DCNN model is required to work accurately in limited computing resources. Meanwhile, the outbreak of the COVID-19 pandemic has led to new challenges in face recognition due to the use of facemasks. Therefore, in this study, we develop a masked face recognition application using a lightweight and efficient DCNN, which is applicable to mobile devices. Two networks for face verification tasks, named MobileFaceNet and SeesawFaceNet are explored for this purpose. We train these models on the augmented version of CelebA dataset, which originally is a set of celebrity images. We put synthetic mask on the face images in CelebA to provide a training dataset contain mix of face images with and without mask. The trained models, which are able to recognize people either wearing or not wearing masks, are then retrained on the face dataset commonly used for verification purposes, i.e. LFW (face images without mask) and MFR2 (face images wearing masks). Transfer learning is utilized to improve the network learning ability, and cosine similarity is adopted to quantify the similarity for pairs of examples. In experiment, the SeesawFaceNet model obtains better performance, with 98.8% accuracy on LFW dataset, 96% accuracy on MFR2 masked dataset. In contrast, the experiment after deployment the models on a smartphone application, the MobileFaceNet model is more superior than the SeesawFaceNet with an accuracy of 85%, an average speed of 44 milliseconds, and model size of 4.9 MB. © 2023 AIP Publishing LLC.

Exploring the Potential of Adopting Computer-graphics Animation to the Switch to a Plant-Based Diet

A healthy lifestyle is a way of living that helps to keep and improve people's health and well-being. Maintaining a healthy lifestyle is essential and a great way to combat many health issues, especially in preparation for the upcoming post-pandemic COVID-19 era. Some studies suggested that animals play a role in spreading SARS-CoV-2, the virus that causes COVID-19, to people. It is thus believed people should be eating more plant-based and less meat diets. Although the development of computer-graphics animation has significantly affected learning media development, there is a lack of evaluative evidence on how it might effectively motivate diet shifts. Therefore, this study explores and analyzes the crucial factors influencing the deliverability and engagement of computer-graphics animation as a learning media to enhance knowledge and promote a plant-based diet. A structured interview, with both closed and open-ended questions, is conducted on various groups of respondents. The analysis results can serve as a guideline for advancing learning media and a basis for future development and research strategies in healthy eating promotion in general. © 2022 IEEE.

Lawyering Social: Navigating Legal Issues on Social Media Posts with a Low-cost Data Algorithm

Along with the increasing use of online communication during the current Covid-19 pandemic, people have become more active in the social media platforms like Facebook and Twitter, whether it be through by posting their content online or by reading the posts of others. Undoubtedly, it has a negative impact in the form of cybercrimes, especially when public understanding of cyber law is still low. To address this issue, we propose an approach to identifying cybercrime on social media posts, with case studies on Indonesian cybercrime law. The main challenge in solving the problem is the use of informal language and non-standard writing. Therefore, we exploit the informative value of terms in that post by paying attention to standard patterns in Indonesian grammar and writing structure with a classifier based on the voting ensemble learning model. The experimental results show that the proposed approach significantly outperformed baselines, with an accuracy of more than 90%. Thus, it proves the effectiveness of the approach in generalizing the content of social media posts according to the category of cybercrime law.

CNN Architecture Comparison for Covid-19 Image Classification Process

Covid-19 detection is the most important stage in the process of diagnosing suspected Covid-19 patients. One of the detections is through lung X-ray images. However, currently, we need an algorithm that can directly detect lung X-ray images that have high accuracy rather than manual detection which has uncertain accuracy. Deep Learning Model using CNN is one way to create the algorithm. In CNN, many architectures can be used, but not all architectures are compatible with the data they have. In this study, we will compare the architectural capabilities of ResNet50, DenseNet121, InceptionV3, VGG16, and MobileNetV2 using 3000 X-ray image data. These research results are that MobileNetV2 gets the highest accuracy value, which is 0.96. Then followed by VGG16 with an accuracy of 0.95, then InceptionV3 with an accuracy of 0.92, followed by DenseNet121 with an accuracy of 0.89, and finally, ResNet50 with an accuracy of 0.86. In the experiment, it was found that the architecture that has a larger number of layers has a lower accuracy value and a higher loss value than the architecture that has a smaller number of layers. © 2022 IEEE.

Detection of Covid-19 Based on Lung Ultrasound Image Using Convolutional Neural Network Architectures

The spread of Covid-19 is so fast that it has become a global pandemic. A fast, cheap, and guaranteed Covid-19 detection system is needed. Medical images such as CT scans and X-rays with biological sciences and deep learning techniques can be critical diagnostic tools. This study uses ultrasound images as an alternative to medical images that can diagnose Covid-19 using a deep learning method based on the Convolutional Neural Network (CNN) architectures. The dataset used is obtained from the Covid-19 Lung Ultrasound. This study shows the highest accuracy of detection covid-19 based on a lung ultrasound image using the VGG16 architecture compared to ResNet50 and InceptionV3architectures. VGG16 architecture with an Adam optimization and a learning rate of 0.0001 yielded 86% accuracy. ResNet50 and InceptionV3architectures produce 79% and 64% of accuracy. © 2021 IEEE.