SemiCurv: Semi-Supervised Curvilinear Structure Segmentation.

Recent work on curvilinear structure segmentation has mostly focused on backbone network design and loss engineering. The challenge of collecting labelled data, an expensive and labor intensive process, has been overlooked. While labelled data is expensive to obtain, unlabelled data is often readily available. In this work, we propose SemiCurv, a semi-supervised learning (SSL) framework for curvilinear structure segmentation that is able to utilize such unlabelled data to reduce the labelling burden. Our framework addresses two key challenges in formulating curvilinear segmentation in a semi-supervised manner. First, to fully exploit the power of consistency based SSL, we introduce a geometric transformation as strong data augmentation and then align segmentation predictions via a differentiable inverse transformation to enable the computation of pixel-wise consistency. Second, the traditional mean square error (MSE) on unlabelled data is prone to collapsed predictions and this issue exacerbates with severe class imbalance (significantly more background pixels). We propose a N-pair consistency loss to avoid trivial predictions on unlabelled data. We evaluate SemiCurv on six curvilinear segmentation datasets, and find that with no more than 5% of the labelled data, it achieves close to 95% of the performance relative to its fully supervised counterpart.

Changes in physical activity, dietary and sleeping pattern among the general population in COVID-19: A systematic review protocol.

BACKGROUND: COVID-19 is a highly infectious respiratory disease caused by a new coronavirus known as SARS-CoV-2. Home confinement and movement restrictions can affect lifestyle changes and may lead to non-communicable diseases (NCD). This systematic review will provide a detailed summary of changing patterns of physical activities, diet and sleep among the general public in COVID-19. METHODS: PubMed, Google Scholar, EMBASE, Science Direct, and Scopus will be, among eight bibliographic databases, applied and search work will take one month (from January 2021 until February 2021). Key search terms will include common characteristics of physical activity, dietary pattern, sleeping pattern, and COVID-19. The reviewers will fully apply the inclusion and exclusion criteria framed by PICOS as well as the screening form and the PRISMA flow for selecting the papers eligible for this review. Moreover, the reviewers will use a self-developed excel table to extract the required information on dietary pattern changes, physical activities and sleep patterns changes, and the Risk of Bias Assessment Tool for Nonrandomized Studies (RoBANS) for practicing quality assessment. We will include only observational studies and analyze the extracted information qualitatively and the review output will be based on the eligible studies' outcomes. DISCUSSION: Changes in physical activity, dietary and sleep patterns are challenging to the public health professionals regarding the risk factors for NCD, and long-term effects might impact the controlling of the NCD. Evidence-based research information is needed regarding the COVID-19 pandemic, and there are a few global data on changes in physical activity, dietary and sleep patterns. Furthermore, innovative public health interventions or implementations are needed to maintain the positive health status of the population in the long run as the consequences of the COVID-19 pandemic. SYSTEMATIC REVIEW REGISTRATION: This systematic review is based on a protocol registered with PROSPERO CRD42021232667.

Polarization Transfer from Ligands Hyperpolarized by Dissolution Dynamic Nuclear Polarization for Screening in Drug Discovery.

Nuclear magnetic resonance (NMR) spectroscopy is a valuable technique for ligand screening, because it exhibits high specificity toward chemical structure and interactions. Dissolution dynamic nuclear polarization (DNP) is a recent advance in NMR methodology that enables the creation of non-equilibrium spin states, which can dramatically increase NMR sensitivity. Here, the transfer of such spin polarization from hyperpolarized ligand to protein is observed. Mixing hyperpolarized benzamidine with the serine protease trypsin, a "fingerprint" of enhanced protein signals is observed, which shows a different intensity profile than the equilibrium NMR spectrum of the protein, but coincides closely to the frequency profile of a saturation transfer difference (STD) NMR experiment. The DNP experiment benefits from hyperpolarization and enables observation of all frequencies in a single, rapid experiment. Based on these merits, it is an interesting alternative to the widely used STD experiment for identification of protein-ligand interactions.