2.5D transfer deep learning model for segmentation of contrast-enhancing lesions on brain magnetic resonance imaging of multiple sclerosis and neuromyelitis optica spectrum disorder.

Background: Multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD) are the two mimic autoimmune diseases of the central nervous system, which are rare in East Asia. Quantitative detection of contrast-enhancing lesions (CELs) on contrast-enhancing T1-weighted magnetic resonance (MR) images is of great significance for assessing the disease activity of MS and NMOSD. However, it is challenging to develop automatic segmentation algorithms due to the lack of data. In this work, we present an automatic segmentation model of CELs based on Fully Convolutional with Attention DenseNet (FCA-DenseNet) and transfer learning strategy to address the challenge of CEL quantification in small-scale datasets. Methods: A transfer learning approach was employed in this study, whereby pretraining was conducted using 77 MS subjects from the open access datasets (MICCAI 2016, MICCAI 2017, ISBI 2015) for white matter hyperintensity segmentation, followed by fine-tuning using 24 MS and NMOSD subjects from the local dataset for CEL segmentation. The proposed FCA-DenseNet combined the Fully Convolutional DenseNet and Convolutional Block Attention Module in order to improve the learning capability. A 2.5D data slicing strategy was used to process complex 3D MR images. U-Net, ResUNet, TransUNet, and Attention-UNet are used as comparison models to FCA-DenseNet. Dice similarity coefficient (DSC), positive predictive value (PPV), true positive rate (TPR), and volume difference (VD) are used as evaluation metrics to evaluate the performances of different models. Results: FCA-DenseNet outperforms all other models in terms of all evaluation metrics, with a DSC of 0.661±0.187, PPV of 0.719±0.201, TPR of 0.680±0.254, and VD of 0.388±0.334. Transfer learning strategy has achieved success in building segmentation models on a small-scale local dataset where traditional deep learning approaches fail to train effectively. Conclusions: The improved FCA-DenseNet, combined with transfer learning strategy and 2.5D data slicing strategy, has successfully addressed the challenges in constructing deep learning models on small-scale datasets, making it conducive to clinical quantification of brain CELs and diagnosis of MS and NMOSD.

Adenosine signalling in T-cell activation favours development of IL-17 positive cells with suppressive properties.

Evidence suggests that the anti-inflammatory nucleoside adenosine can shape immune responses by shifting the regulatory (Treg )/helper (Th17) T-cell balance in favour of Treg . Since this observation is based on in vivo and in vitro studies mostly confined to murine models, we comprehensively analysed effects of adenosine on human T-cells. Proliferation, phenotype and cytokine production of stimulated T-cells were assessed by flow cytometry, multiplex assay and ELISA, gene expression profiling was determined by microarray. We found that the pan-adenosine agonist 5'-N-ethylcarboxamidoadenosine (NECA) skews human CD3+ T-cell responses towards non-inflammatory Th17 cells. Addition of NECA during T-cell activation increased the development of IL-17+ cells with a CD4+ RORγt+ phenotype and enhanced CD161 and CD196 surface expression. Remarkably, these Th17 cells displayed non-inflammatory cytokine and gene expression profiles including reduced Th1/Th17 transdifferentiation, a stem cell-like molecular signature and induced surface expression of the adenosine-producing ectoenzymes CD39 and CD73. Thus, T-cells cultured under Th17-inducing conditions together with NECA were capable of suppressing responder T-cells. Finally, genome-wide gene expression profiling revealed metabolic quiescence previously associated with non-pathogenic Th17 cells in response to adenosine signalling. Our data suggest that adenosine induces non-inflammatory Th17 cells in human T-cell differentiation, potentially through regulation of metabolic pathways.

AI support for accurate and fast radiological diagnosis of COVID-19: an international multicenter, multivendor CT study.

OBJECTIVES: Differentiation between COVID-19 and community-acquired pneumonia (CAP) in computed tomography (CT) is a task that can be performed by human radiologists and artificial intelligence (AI). The present study aims to (1) develop an AI algorithm for differentiating COVID-19 from CAP and (2) evaluate its performance. (3) Evaluate the benefit of using the AI result as assistance for radiological diagnosis and the impact on relevant parameters such as accuracy of the diagnosis, diagnostic time, and confidence. METHODS: We included n = 1591 multicenter, multivendor chest CT scans and divided them into AI training and validation datasets to develop an AI algorithm (n = 991 CT scans; n = 462 COVID-19, and n = 529 CAP) from three centers in China. An independent Chinese and German test dataset of n = 600 CT scans from six centers (COVID-19 / CAP; n = 300 each) was used to test the performance of eight blinded radiologists and the AI algorithm. A subtest dataset (180 CT scans; n = 90 each) was used to evaluate the radiologists' performance without and with AI assistance to quantify changes in diagnostic accuracy, reporting time, and diagnostic confidence. RESULTS: The diagnostic accuracy of the AI algorithm in the Chinese-German test dataset was 76.5%. Without AI assistance, the eight radiologists' diagnostic accuracy was 79.1% and increased with AI assistance to 81.5%, going along with significantly shorter decision times and higher confidence scores. CONCLUSION: This large multicenter study demonstrates that AI assistance in CT-based differentiation of COVID-19 and CAP increases radiological performance with higher accuracy and specificity, faster diagnostic time, and improved diagnostic confidence. KEY POINTS: • AI can help radiologists to get higher diagnostic accuracy, make faster decisions, and improve diagnostic confidence. • The China-German multicenter study demonstrates the advantages of a human-machine interaction using AI in clinical radiology for diagnostic differentiation between COVID-19 and CAP in CT scans.

Concurrent stimulation of monocytes with CSF1 and polarizing cytokines reveals phenotypic and functional differences with classical polarized macrophages.

In atherosclerotic lesions, macrophages are exposed to CSFs and various microenvironmental cues, which ultimately drive their polarization state. We studied the expression of different CSFs in artery specimen and cultured vascular cells and assessed whether concurrent stimulation (CS) of monocytes with CSF1 and polarizing cytokines generated macrophages (CSM1 and CSM2) that were phenotypically and functionally different from classically polarized M1 and M2 macrophages. We also assessed the influence of acetylsalicylic acid (ASA) on the capacity of polarized macrophages to stimulate T-cell proliferation. CSF1 was the most prominent CSF expressed in arteries and cultured vascular cells. M1 and CSM1 macrophages differed in CD86 and CD14 expression, which was up-regulated respectively down-regulated by LPS. M2 and CSM2 macrophages were phenotypically similar. Cyclooxygenase expression was different in CSM1 (COX-1- and COX-2+ after LPS stimulation) and CSM2 (COX-1+ and COX-2- ) macrophages. TNFα production was more pronounced in CSM1 macrophages, whereas IL-10 was produced at higher levels by CSM2 macrophages. Proliferation of allogeneic T cells was strongly supported by CSM2, but not by CSM1 polarized macrophages. Although ASA did not affect anti-CD3/CD28-mediated proliferation, it significantly reduced CSM2 and CSM1-mediated T-cell proliferation. Supernatants of LPS-stimulated CSM2 but not of CSM1 macrophages could overcome the inhibition by ASA. Hence, we demonstrate that CSM1 and CSM2 macrophages are phenotypically and to some extent functionally distinct from classically polarized M1 and M2 macrophages. CSM2 macrophages produce a COX-1-dependent soluble factor that supports T-cell proliferation, the identity hereof is still elusive and warrants further studies.

3D-QSAR and docking studies of 3-arylquinazolinethione derivatives as selective estrogen receptor modulators.

3D-QSAR and molecular docking analysis were performed to explore the interaction of estrogen receptors (ERalpha and ERbeta) with a series of 3-arylquinazolinethione derivatives. Using the conformations of these compounds revealed by molecular docking, CoMFA analysis resulted in the first quantitative structure-activity relationship (QSAR) and first quantitative structure-selectivity relationship (QSSR) models predicting the inhibitory activity against ERbeta and the selectivity against ERá. The q(2) and R(2) values, along with further testing, indicate that the obtained 3D-QSAR and 3D-QSSR models will be valuable in predicting both the inhibitory activity and selectivity of 3-arylquinazolinethione derivatives for these protein targets. A set of 3D contour plots drawn based on the 3D-QSAR and 3D-QSSR models reveal modifications of substituents at C2 and C5 of the quinazoline which my be useful to improve both the activity and selectivity of ERbeta/ ERalpha. Results showed that both the steric and electrostatic factors should appropriately be taken into account in future rational design and development of more active and more selective ERbeta inhibitors for the therapeutic treatment of osteoporosis.

3D-QSAR study of hallucinogenic phenylalkylamines by using CoMFA approach.

The three-dimensional quantitative structure-activity relationship (3D-QSAR) has been studied on 90 hallucinogenic phenylalkylamines by the comparative molecular field analysis (CoMFA). Two conformations were compared during the modeling. Conformation I referred to the amino group close to ring position 6 and conformation II related to the amino group trans to the phenyl ring. Satisfactory results were obtained by using both conformations. There were still differences between the two models. The model based on conformation I got better statistical results than the one about conformation II. And this may suggest that conformation I be preponderant when the hallucinogenic phenylalkylamines interact with the receptor. To further confirm the predictive capability of the CoMFA model, 18 compounds with conformation I were randomly selected as a test set and the remaining ones as training set. The best CoMFA model based on the training set had a cross-validation coefficient q (2) of 0.549 at five components and non cross-validation coefficient R (2) of 0.835, the standard error of estimation was 0.219. The model showed good predictive ability in the external test with a coefficient R (pre) (2) of 0.611. The CoMFA coefficient contour maps suggested that both steric and electrostatic interactions play an important role. The contributions from the steric and electrostatic fields were 0.450 and 0.550, respectively.