1221 Deciphering cellular features: StyleGAN for enhancing H&E cell classification

Tumor-specific T cells act as major actors in underpinning effective immune checkpoint-blockade therapies in cancer, but most of the studies focus on CD8+ tumor-infiltrating lymphocytes (TILs). Although roles of CD4+ T cells and especially Tregs in cancer have been established, the basis for their phenotypic heterogeneity in human tumor infiltrates is not well determined. Here we profile ex-vivounexpanded CD4+ TILs in human lung and colorectal cancer using single-cell and mass-cytometry analysis. Surprisingly, up to 80% of CD4+ TILs are composed of non-Treg cells (FoxP3−). Non-Treg CD4+ TILs are highly heterogeneous and consist of different populations of effector (CD45RO+), cytotoxic (Granzyme B+), senescent (CD57+) and exhausted cells at different levels (PD-1+/++, CTLA-4+/−, TIGIT+, CD39+/−). Recently, we showed that the surface marker CD39 can accurately discriminate between bystander (CD39−) and tumor-specific (CD39+) CD8+ TILs. Across all tumors tested, more than 90% of Tregs cells expressed CD39. However, frequencies of CD39+ among non-Treg CD4+ TILs varied dramatically from 0% to 90% and strongly correlated (r2=0.8) with the frequency of CD39+CD8+ TILs in our cohort of patients. These observations suggest the presence of bystander CD4+ TILs and a high degree of coordination between the tumor-specific CD4 and CD8 T cell responses. Using in-vitroassay, we aim to confirm the presence of these bystander CD4+ TILs specific for cancer-unrelated antigens (HCMV, EBV, Influenza). Overall, our findings could highlight that not all CD4+ TILs are tumor-specific and could suggest measuring CD39 expression as a straightforward way to quantify or isolate tumor-specific T cells, thus opening new diagnostic and therapeutic avenues.

Culturing cells in three-dimensions (3D) has become a well established approach as it is more representative of the in vivo environment than traditional two-dimensional (2D) cultures. Allowing cells to interact with each other in a spheroid creates a micro-environment which mimics in vivo tissue and is a better model for examining the effect of drugs in cancer. Developing uniform spheroids becomes especially important as it forms the basis for robust and reliable assays. S-BIO PrimeSurface® cultureware are ultra low attachment (ULA) dishes and plates that promote scaffold free, self assembly spheroid formation. The plates are pre-coated with a proprietary hydrophilic polymer that enables spontaneous spheroid formation of uniform size. PrimeSurface 96 and 384 ULA plates have good optical clarity making them highly suitable for brightfield and fluorescent imaging. Imaging technologies such as the BioTek Cytation™ 5 allows researchers to study not only spheroid proliferation through brightfield imaging, but also phenotypic events such as hypoxia, apoptosis, or necrosis induction through the use of fluorescent probes and fluorescence imaging. Incorporation of z-stacking and projection techniques in the Gen5™ Microplate Reader and Imager Software create in-focus images of spheroidal cells, allowing accurate, robust, and repeatable determination of the effect of test molecules or conditions. In this app note, we present data generated with BioTek Cytation 5 using PrimeSurface ULA plates to develop simple robust spheroid assays for brightfield and fluorescence imaging.

Quantitative analysis of cell structures is essential for biomedical and pharmaceutical research. The standard imaging approach relies on fluorescence microscopy, where cell structures of interest are labeled by chemical staining techniques. However, these techniques are often invasive and sometimes even toxic to the cells, in addition to being time consuming, labor intensive, and expensive. Here, we introduce an alternative deep-learning–powered approach based on the analysis of bright-field images by a conditional generative adversarial neural network (cGAN). We show that this is a robust and fast-converging approach to generate virtually stained images from the bright-field images and, in subsequent downstream analyses, to quantify the properties of cell structures. Specifically, we train a cGAN to virtually stain lipid droplets, cytoplasm, and nuclei using bright-field images of human stem-cell–derived fat cells (adipocytes), which are of particular interest for nanomedicine and vaccine development. Subsequently, we use these virtually stained images to extract quantitative measures about these cell structures. Generating virtually stained fluorescence images is less invasive, less expensive, and more reproducible than standard chemical staining; furthermore, it frees up the fluorescence microscopy channels for other analytical probes, thus increasing the amount of information that can be extracted from each cell. To make this deep-learning–powered approach readily available for other users, we provide a Python software package, which can be easily personalized and optimized for specific virtual-staining and cell-profiling applications.

This article addresses the semantic segmentation of laparoscopic surgery images, placing special emphasis on the segmentation of structures with a smaller number of observations. As a result of this study, adjustment parameters are proposed for deep neural network architectures, enabling a robust segmentation of all structures in the surgical scene. The U-Net architecture with five encoder-decoders (U-Net5ed), SegNet-VGG19, and DeepLabv3+ employing different backbones are implemented. Three main experiments are conducted, working with Rectified Linear Unit (ReLU), Gaussian Error Linear Unit (GELU), and Swish activation functions. The applied loss functions include Cross Entropy (CE), Focal Loss (FL), Tversky Loss (TL), Dice Loss (DiL), Cross Entropy Dice Loss (CEDL), and Cross Entropy Tversky Loss (CETL). The performance of Stochastic Gradient Descent with momentum (SGDM) and Adaptive Moment Estimation (Adam) optimizers is compared. It is qualitatively and quantitatively confirmed that DeepLabv3+ and U-Net5ed architectures yield the best results. The DeepLabv3+ architecture with the ResNet-50 backbone, Swish activation function, and CETL loss function reports a Mean Accuracy (MAcc) of 0.976 and Mean Intersection over Union (MIoU) of 0.977. The semantic segmentation of structures with a smaller number of observations, such as the hepatic vein, cystic duct, Liver Ligament, and blood, verifies that the obtained results are very competitive and promising compared to the consulted literature. The proposed selected parameters were validated in the YOLOv9 architecture, which showed an improvement in semantic segmentation compared to the results obtained with the original architecture.

Immune Regulatory Mechanisms Influence Early Pathology in Spinal Cord Injury and in Spontaneous Autoimmune Encephalomyelitis

Tropism of Severe Acute Respiratory Syndrome Coronavirus 2 for Barrett’s Esophagus May Increase Susceptibility to Developing Coronavirus Disease 2019

Cervical cancer is one of the most common types of cancer among women, which has higher death-rate than many other cancer types. The most common way to diagnose cervical cancer is to analyze images of cervical cells, which is performed using Pap smear imaging test. Early and accurate diagnosis can save the lives of many patients and increase the chance of success of treatment methods. Until now, various methods have been proposed to diagnose cervical cancer based on the analysis of Pap smear images. Most of the existing methods can be divided into two groups of methods based on deep learning techniques or machine learning algorithms. In this study, a combination method is presented, whose overall structure is based on a machine learning strategy, where the feature extraction stage is completely separate from the classification stage. However, in the feature extraction stage, deep networks are used. In this paper, a multi-layer perceptron (MLP) neural network fed with deep features is presented. The number of hidden layer neurons is tuned based on four innovative ideas. Additionally, ResNet-34, ResNet-50 and VGG-19 deep networks have been used to feed MLP. In the presented method, the layers related to the classification phase are removed in these two CNN networks, and the outputs feed the MLP after passing through a flatten layer. In order to improve performance, both CNNs are trained on related images using the Adam optimizer. The proposed method has been evaluated on the Herlev benchmark database and has provided 99.23 percent accuracy for the two-classes case and 97.65 percent accuracy for the 7-classes case. The results have shown that the presented method has provided higher accuracy than the baseline networks and many existing methods.

Thermo-diffusion and non-uniform heat source/sink effects on hydromagnetic flow of Cu and TiO2 water - based nanofluid partially filled with a porous medium

Semiautomated Multiplexed Quantum Dot-Based in Situ Hybridization and Spectral Deconvolution

Light scattering imaging modal expansion cytometry for label-free single-cell analysis with deep learning.

Multifunctional probes with high MRI sensitivity and high efficiency for cell labeling are desirable for MR cell imaging. Herein, we have fabricated fluorescent mesoporous silica-coated superparamagnetic iron oxide nanoparticles (fmSiO4@SPIONs) for neural progenitor cell (C17.2) MR imaging. FmSiO4@SPIONs were discrete and uniform in size, and had a clear core-shell structure. The magnetic core size was about 10 nm and the fluorescent mesoporous silica coating layer was around 20 nm. Compared with fluorescent dense silica-coated SPIONs (fdSiO4@SPIONs) with a similar size, fmSiO4@SPIONs demonstrated higher MR sensitivity and cell labeling efficiency. When implanted into the right hemisphere of stroke mice, contralateral to the ischemic territory, a small amount of labeled cells were able to be tracked migrating to the lesion sites using a clinical MRI scanner (3 T). More impressively, even when administered intravenously, the labeled cells could also be monitored homing to the ischemic area. MRI observations were corroborated by histological studies of the brain tissues. Our study demonstrated that fmSiO4@SPIONs are highly effective for cell imaging and hold great promise for MRI cell tracking in future.

HaloTag protein-mediated specific labeling of living cells with quantum dots

The study of embryonic events using different animal model systems is crucial for gaining insights into human development and birth defects. Biological imaging plays a major role in this effort by providing a spatiotemporal framework to link complex cell movements with molecular data. However, depending on the age of the embryo and the location of a morphogenetic event, visualization often requires the design of novel culture and imaging techniques. One of the primary model systems for biological imaging is the avian embryo, due to its accessibility to manipulation, relatively two-dimensional morphogenesis early on, and viability when grown in culture. Significant work in avian embryo culture and cell labeling, together with advances in imaging technology, now make it possible to monitor many developmental events within the period from egg laying to hatching. Here, we present the latest in avian developmental imaging, focusing on cell labeling, embryo culture, and imaging technologies.

Manganese serves as a cofactor to a variety of proteins necessary for proper bodily development and function. However, an overabundance of Mn in the brain can result in manganism, a neurological condition resembling Parkinson's disease (PD). Bulk sample measurement techniques have identified the globus pallidus and thalamus as targets of Mn accumulation in the brain, however smaller structures/cells cannot be measured. Here, X-ray fluorescence microscopy determined the metal content and distribution in the substantia nigra (SN) of the rodent brain. In vivo retrograde labeling of dopaminergic cells (via FluoroGold™) of the SN pars compacta (SNc) subsequently allowed for XRF imaging of dopaminergic cells in situ at subcellular resolution. Chronic Mn exposure resulted in a significant Mn increase in both the SN pars reticulata (>163%) and the SNc (>170%) as compared to control; no other metal concentrations were significantly changed. Subcellular imaging of dopaminergic cells demonstrated that Mn is located adjacent to the nucleus. Measured intracellular manganese concentrations range between 40-200 μM; concentrations as low as 100 μM have been observed to cause cell death in cell cultures. Direct observation of Mn accumulation in the SNc could establish a biological basis for movement disorders associated with manganism, specifically Mn caused insult to the SNc. Accumulation of Mn in dopaminergic cells of the SNc may help clarify the relationship between Mn and the loss of motor skills associated with manganism.

A novel type of bimodal imaging nanoprobe based on (19)F-based magnetic resonance imaging and near-infrared optical imaging has been synthesized and applied for the labeling and imaging of dendritic cells both in vitro and in vivo.