Masked Face Recognition Using Histogram-Based Recurrent Neural Network.

Masked face recognition (MFR) is an interesting topic in which researchers have tried to find a better solution to improve and enhance performance. Recently, COVID-19 caused most of the recognition system fails to recognize facial images since the current face recognition cannot accurately capture or detect masked face images. This paper introduces the proposed method known as histogram-based recurrent neural network (HRNN) MFR to solve the undetected masked face problem. The proposed method includes the feature descriptor of histograms of oriented gradients (HOG) as the feature extraction process and recurrent neural network (RNN) as the deep learning process. We have proven that the combination of both approaches works well and achieves a high true acceptance rate (TAR) of 99 percent. In addition, the proposed method is designed to overcome the underfitting problem and reduce computational burdens with large-scale dataset training. The experiments were conducted on two benchmark datasets which are RMFD (Real-World Masked Face Dataset) and Labeled Face in the Wild Simulated Masked Face Dataset (LFW-SMFD) to vindicate the viability of the proposed HRNN method.

Insight into free energy and dynamic cross-correlations of residue for binding affinity of antibody and receptor binding domain SARS-CoV-2.

SARS-CoV-2 virus continues to evolve and mutate causing most of the mutated variants resist to many of the therapeutic monoclonal antibodies (mAbs). Despite several mAbs retained neutralizing capability for Omicron BA.1 and BA.2, reduction in neutralization potency was reported. Hence, effort of searching for mAb that is broader in neutralization breadth without losing the neutralizing ability is continued. MW06 was reported with capability in neutralizing most of the variants of concern (VOC) and it binds to the conserved region (left flank) near epitope mAb sotrovimab (S309). In this study, binding affinity of mAb MW06 and its cocktail formulation with MW05 for receptor binding domain (RBD) SARS-CoV-2 virus was investigated under molecular dynamics simulations (MDs). Binding free energies computed by Molecular Mechanics Generalised Born Surface Area (MM-GBSA) algorithm predicted the binding affinity of MW06 for RBD BA.1 (-53 kcal/mol) as strong as RBD wildtype (-58 kcal/mol) while deterioration was observed for RBD BA.2 (-43 kcal/mol). Alike S309 and MW06, simulated cocktail mAb (MW05 and MW06)-RBD interactions suggested the neutralizing capability of the cocktail formulation for RBD BA.1 and BA.2 reduced. Meanwhile, residue pairs that favour the communication between the mAb and RBD have been identified by decomposing the free energy per pairwise residue basis. Apart from understanding the effects of mutation occurred in the RBD region on human angiotensin-converting enzyme 2 (hACE2) binding, impact of heavily mutated RBD on mAb-RBD interactions was investigated in this study as well. In addition to energetic profile obtained from MDs, plotting the dynamics cross-correlation map of the mAb-RBD complex under elastic network model (ENM) was aimed to understand the cross-correlations between residue fluctuations. It allows simple and rapid analysis on the motions or dynamics of the protein residues of mAbs and RBD in complex. Protein residues having correlated motions are normally part of the structural domains of the protein and their respective motions and protein function are related. Motion of mutated RBD residues and mAb residues was less correlated while their respective interactions energy computed to be higher. The combined techniques of MDs and ENM offered simplicity in understanding dynamics and energy contribution that explain binding affinity of mAb-RBD complexes.

Masked face recognition with principal random forest convolutional neural network (PRFCNN)

Masked face recognition embarks the interest among the researchers to find a better algorithm to improve the performance of face recognition applications, especially in the Covid-19 pandemic lately. This paper introduces a proposed masked face recognition method known as Principal Random Forest Convolutional Neural Network (PRFCNN). This method utilizes the strengths of Principal Component Analysis (PCA) with the combination of Random Forest algorithm in Convolution Neural Network to pre-train the masked face features. PRFCNN is designed to assist in extracting more salient features and prevent overfitting problems. Experiments are conducted on two benchmarked datasets, RMFD (Real-World Masked Face Dataset) and LFW Simulated Masked Face Dataset using various parameter settings. The experimental result with a minimum recognition rate of 90% accuracy promises the effectiveness of the proposed PRFCNN over the other state-of-the-art methods. [ FROM AUTHOR]

Chest lesion CT radiological features and quantitative analysis in RT-PCR turned negative and clinical symptoms resolved COVID-19 patients.

BACKGROUND: Many studies have described lung lesion computed tomography (CT) features of coronavirus disease 2019 (COVID-19) patients at the early and progressive stages. In this study, we aim to evaluate lung lesion CT radiological features along with quantitative analysis for the COVID-19 patients ready for discharge. METHODS: From February 10 to March 10, 2020, 125 COVID-19 patients (age: 16-67 years, 63 males) ready for discharge, with two consecutive negative reverse transcription-polymerase chain reaction (RT-PCR) and no clinical symptoms for more than 3 days, were included. The pre-discharge CT was performed on all patients 1-3 days after the second negative RT-PCR test, and the follow-up CTs were performed on 44 patients 2-13 days later. The imaging features and quantitative analysis were evaluated on both the pre-discharge and the follow-up CTs, by both radiologists and an artificial intelligence (AI) software. RESULTS: On the pre-discharge CT, the most common CT findings included ground-glass opacity (GGO) (99/125, 79.2%) with bilateral mixed distribution, and fibrosis (56/125, 44.8%) with bilateral subpleural distribution. Enlarged mediastinal lymph nodes were also commonly observed (45/125, 36.0%). AI enabled quantitative analysis showed the right lower lobe was mostly involved, and lesions most commonly had CT value of -570 to -470 HU consistent with GGO. Follow-up CT showed GGO decrease in size and density (40/40, 100%) and fibrosis reduction (17/26, 65.4%). Compared with the pre-discharge CT results, quantitative analysis shows the lung lesion volume regressed significantly at follow-up. CONCLUSIONS: For COVID-19 patients ready for discharge, GGO and fibrosis are the main CT features and they further regress at follow-up.