Querying Labeled for Unlabeled: Cross-Image Semantic Consistency Guided Semi-Supervised Semantic Segmentation.

Semi-supervised semantic segmentation aims to learn a semantic segmentation model via limited labeled images and adequate unlabeled images. The key to this task is generating reliable pseudo labels for unlabeled images. Existing methods mainly focus on producing reliable pseudo labels based on the confidence scores of unlabeled images while largely ignoring the use of labeled images with accurate annotations. In this paper, we propose a Cross-Image Semantic Consistency guided Rectifying (CISC-R) approach for semi-supervised semantic segmentation, which explicitly leverages the labeled images to rectify the generated pseudo labels. Our CISC-R is inspired by the fact that images belonging to the same class have a high pixel-level correspondence. Specifically, given an unlabeled image and its initial pseudo labels, we first query a guiding labeled image that shares the same semantic information with the unlabeled image. Then, we estimate the pixel-level similarity between the unlabeled image and the queried labeled image to form a CISC map, which guides us to achieve a reliable pixel-level rectification for the pseudo labels. Extensive experiments on the PASCAL VOC 2012, Cityscapes, and COCO datasets demonstrate that the proposed CISC-R can significantly improve the quality of the pseudo labels and outperform the state-of-the-art methods. Code is available at https://github.com/Luffy03/CISC-R.

Self-supervised patient-specific features learning for OCT image classification.

Deep learning's great success in image classification is heavily reliant on large-scale annotated datasets. However, obtaining labels for optical coherence tomography (OCT) data requires the significant effort of professional ophthalmologists, which hinders the application of deep learning in OCT image classification. In this paper, we propose a self-supervised patient-specific features learning (SSPSF) method to reduce the amount of data required for well OCT image classification results. Specifically, the SSPSF consists of a self-supervised learning phase and a downstream OCT image classification learning phase. The self-supervised learning phase contains two self-supervised patient-specific features learning tasks. One is to learn to discriminate an OCT scan which belongs to a specific patient. The other task is to learn the invariant features related to patients. In addition, our proposed self-supervised learning model can learn inherent representations from the OCT images without any manual labels, which provides well initialization parameters for the downstream OCT image classification model. The proposed SSPSF achieves classification accuracy of 97.74% and 98.94% on the public RETOUCH dataset and AI Challenger dataset, respectively. The experimental results on two public OCT datasets show the effectiveness of the proposed method compared with other well-known OCT image classification methods with less annotated data.

Intra- and Inter-Slice Contrastive Learning for Point Supervised OCT Fluid Segmentation.

OCT fluid segmentation is a crucial task for diagnosis and therapy in ophthalmology. The current convolutional neural networks (CNNs) supervised by pixel-wise annotated masks achieve great success in OCT fluid segmentation. However, requiring pixel-wise masks from OCT images is time-consuming, expensive and expertise needed. This paper proposes an Intra- and inter-Slice Contrastive Learning Network (ISCLNet) for OCT fluid segmentation with only point supervision. Our ISCLNet learns visual representation by designing contrastive tasks that exploit the inherent similarity or dissimilarity from unlabeled OCT data. Specifically, we propose an intra-slice contrastive learning strategy to leverage the fluid-background similarity and the retinal layer-background dissimilarity. Moreover, we construct an inter-slice contrastive learning architecture to learn the similarity of adjacent OCT slices from one OCT volume. Finally, an end-to-end model combining intra- and inter-slice contrastive learning processes learns to segment fluid under the point supervision. The experimental results on two public OCT fluid segmentation datasets (i.e., AI Challenger and RETOUCH) demonstrate that the ISCLNet bridges the gap between fully-supervised and weakly-supervised OCT fluid segmentation and outperforms other well-known point-supervised segmentation methods.

Structure-Guided Cross-Attention Network for Cross-Domain OCT Fluid Segmentation.

Accurate retinal fluid segmentation on Optical Coherence Tomography (OCT) images plays an important role in diagnosing and treating various eye diseases. The art deep models have shown promising performance on OCT image segmentation given pixel-wise annotated training data. However, the learned model will achieve poor performance on OCT images that are obtained from different devices (domains) due to the domain shift issue. This problem largely limits the real-world application of OCT image segmentation since the types of devices usually are different in each hospital. In this paper, we study the task of cross-domain OCT fluid segmentation, where we are given a labeled dataset of the source device (domain) and an unlabeled dataset of the target device (domain). The goal is to learn a model that can perform well on the target domain. To solve this problem, in this paper, we propose a novel Structure-guided Cross-Attention Network (SCAN), which leverages the retinal layer structure to facilitate domain alignment. Our SCAN is inspired by the fact that the retinal layer structure is robust to domains and can reflect regions that are important to fluid segmentation. In light of this, we build our SCAN in a multi-task manner by jointly learning the retinal structure prediction and fluid segmentation. To exploit the mutual benefit between layer structure and fluid segmentation, we further introduce a cross-attention module to measure the correlation between the layer-specific feature and the fluid-specific feature encouraging the model to concentrate on highly relative regions during domain alignment. Moreover, an adaptation difficulty map is evaluated based on the retinal structure predictions from different domains, which enforces the model focus on hard regions during structure-aware adversarial learning. Extensive experiments on the three domains of the RETOUCH dataset demonstrate the effectiveness of the proposed method and show that our approach produces state-of-the-art performance on cross-domain OCT fluid segmentation.

Multi-Modal Retinal Image Classification With Modality-Specific Attention Network.

Recently, automatic diagnostic approaches have been widely used to classify ocular diseases. Most of these approaches are based on a single imaging modality (e.g., fundus photography or optical coherence tomography (OCT)), which usually only reflect the oculopathy to a certain extent, and neglect the modality-specific information among different imaging modalities. This paper proposes a novel modality-specific attention network (MSAN) for multi-modal retinal image classification, which can effectively utilize the modality-specific diagnostic features from fundus and OCT images. The MSAN comprises two attention modules to extract the modality-specific features from fundus and OCT images, respectively. Specifically, for the fundus image, ophthalmologists need to observe local and global pathologies at multiple scales (e.g., from microaneurysms at the micrometer level, optic disc at millimeter level to blood vessels through the whole eye). Therefore, we propose a multi-scale attention module to extract both the local and global features from fundus images. Moreover, large background regions exist in the OCT image, which is meaningless for diagnosis. Thus, a region-guided attention module is proposed to encode the retinal layer-related features and ignore the background in OCT images. Finally, we fuse the modality-specific features to form a multi-modal feature and train the multi-modal retinal image classification network. The fusion of modality-specific features allows the model to combine the advantages of fundus and OCT modality for a more accurate diagnosis. Experimental results on a clinically acquired multi-modal retinal image (fundus and OCT) dataset demonstrate that our MSAN outperforms other well-known single-modal and multi-modal retinal image classification methods.

Septin4_i1 regulates apoptosis in hepatic stellate cells through peroxisome proliferator-activated receptor-γ/Akt/B-cell lymphoma 2 pathway.

Apoptosis of activated hepatic stellate cells (HSCs) has been verified as a potential mechanism to aid in hepatic fibrosis remission. Earlier research suggests that Septin4_i1 may sensitize hepatocellular carcinoma cells to serum starvation-induced apoptosis. Here, we aimed to investigate the effect of Septin4_i1 on HSC apoptosis and explore the associated signaling pathways. We found that Septin4_i1 can induce apoptosis in LX-2 cells and that this is accompanied by an up-regulation in cleaved-caspase-3 and peroxisome proliferator-activated receptor-γ (PPAR-γ) expression and a down-regulation in α-SMA expression. Over-expression of Septin4_i1 reduced phosphorylated Akt and B-cell lymphoma 2 (Bcl-2) expression but had no effect on the expression of p53 and death receptor (DR)-5. The decreased expression of Bcl-2 and the increased expression of cleaved-caspase-3 induced by Sept4_i1 could be reversed by GW501516, a PPAR-ß/δ agonist that has been reported by others to enhance Akt signaling. In addition, GW9662, an antagonist of PPAR-γ, could also inhibit apoptosis in LX-2 cells induced by Sept4_i1. In conclusion, our data suggest that Sept4_i1 induces HSC apoptosis by inhibiting Akt and Bcl-2 expression and up-regulating PPAR-γ expression.

Schistosoma japonicum soluble egg antigens induce apoptosis and inhibit activation of hepatic stellate cells: a possible molecular mechanism.

Hepatic stellate cells play a key role in the development of hepatic fibrosis. Activated hepatic stellate cells can be reversed to a quiescent-like state or apoptosis can be induced to reverse fibrosis. Some studies have recently shown that Schistosoma mansoni eggs could suppress the activation of hepatic stellate cells and that soluble egg antigens from schistosome eggs could promote immunocyte apoptosis. Hence, in this study, we attempt to assess the direct effects of Schistosoma japonicum soluble egg antigens on hepatic stellate cell apoptosis, and to explore the mechanism by which the apoptosis of activated hepatic stellate cells can be induced by soluble egg antigens, as well as the mechanism by which hepatic stellate cell activation is inhibited by soluble egg antigens. Here, it was shown that S. japonicum-infected mouse livers had increased apoptosis phenomena and a variability of peroxisome proliferator-activated receptor γ expression. Soluble egg antigens induce morphological changes in the hepatic stellate cell LX-2 cell line, inhibit cell proliferation and induce cell-cycle arrest at the G1 phase. Soluble egg antigens also induce apoptosis in hepatic stellate cells through the TNF-related apoptosis-inducing ligand/death receptor 5 and caspase-dependent pathways. Additionally, soluble egg antigens could inhibit the activation of hepatic stellate cells through peroxisome proliferator-activated receptor γ and the transforming growth factor ß signalling pathways. Therefore, our study provides new insights into the anti-fibrotic effects of S. japonicum soluble egg antigens on hepatic stellate cell apoptosis and the underlying mechanism by which the liver fibrosis could be attenuated by soluble egg antigens.

Expression of Septin4 in human hepatic stellate cells LX-2 stimulated by LPS.

Septin4, a member of polymerizing GTP-binding proteins family, is reported to be involved in cytoskeletal organization in mitosis, apoptosis, fibrosis, and other cellular processes. Since various Septin4 expression patterns were reported in different diseases, this study aimed to investigate Septin4 expression in human LX-2 cell line stimulated by lipopolysaccharides (LPS) and attempted to clarify the relationship between Septin4 and hepatic inflammatory injury and fibrosis. In this subject, human stellate cell line LX-2 was stimulated by LPS. The expression of Septin4 was analyzed by Western blot and quantitative real-time PCR. To observe the relationship among Toll-like receptor 4 (TLR4), TGF-ß, and Septin4, proteins from the anti-TLR4 antibody blocked cells, as well as the TGF-ß-induced cells, were analyzed by the method of Western blot. As the results, LPS could induce the alteration of α-smooth muscle actin and Septin4 expression in LX-2 cells. Septin4 expression was regulated by LPS stimulation through TLR4 and TGF-ß pathway. These results therefore suggest that Septin4 may be involved in the process of activation of hepatic stellate cells by LPS stimulation. Further work would focus on the function of Septin4 in hepatic inflammatory injury and fibrosis.

ß-1,4-Galactosyltransferase-involved in lipopolysaccharide-induced adhesion of Schwann cells.

OBJECTIVE AND DESIGN: The carbohydrate moieties of glycoprotein are associated with some inflammatory diseases by affecting a wide range of biological functions of cells. This study aimed to investigate the role of ß-1,4-galactosyltransferase-I (ß-1,4-GalT-I) in adhesion of Schwann cells during inflammation. SUBJECTS: A rat Schwann cell line, RSC 96 was used. METHODS: We used western blotting to detect the expression of ß-1,4-GalT-I. Flow cytomety was used to measure the galactosylation of glycoproteins on cell surfaces. Immunofluorescent staining was used to examine the expression of α6 integrin, focal adhesion kinase (FAK) and F-actin. Tyrosine phosphorylation of FAK was detected by immunoprecipitation. An adhesion assay was performed to investigate the adhesion of Schwann cells. One-way ANOVA was used to compare differences between the operated and the control group. RESULTS: Schwann cell adhesion was induced by LPS stimulation and was accompanied by upregulation of ß-1,4-GalT-I expression and galactosylation of glycoproteins. There was a change of localization of FAK and cytoskeleton organization in LPS treated cells compared with control cells. The pretreated cells enhanced tyrosine phosphorylation of FAK compared with control cells in the adhesion process. With the increased cell surface expression of α6 integrin and ß-1,4-GalT-I, the adhesion of Schwann cells on laminin was increased as well. CONCLUSIONS: These results suggested that ß-1,4-GalT-I may play an important role in adhesion of Schwann cells during inflammation.

Involvement of Src-suppressed C kinase substrate in experimental autoimmune encephalomyelitis: a link between release of astrocyte proinflammatory factor and oligodendrocyte apoptosis.

Src-suppressed C kinase substrate (SSeCKS) is involved in inflammation in the central nervous system (CNS), and plays a role in control of cell signaling and cytoskeletal arrangement. However, the expression and function of SSeCKS and its function in multiple sclerosis (MS) and its common animal model, experimental autoimmune encephalomyelitis (EAE) remained to be elucidated. In the present study, we first reported that SSeCKS was remarkably increased in astrocytes of EAE rats in vivo. TNF-alpha and NO were significantly induced in astrocytes stimulated with LPS/IFN-gamma in vitro, which was blocked in astrocytes transfected with SSeCKS siRNA. These results indicated that SSeCKS played a role in the production of TNF-alpha and NO in astrocytes with inflammatory stimulation. As excessive release of TNF-alpha and NO were major mediators in autoimmune diseases and correlated with oligodendrocyte cell death, we further investigated whether SSeCKS participated in oligodendrocyte apoptosis. Conditioned media (CM) from astrocytes treated with LPS/IFN-gamma decreased oligodendrocyte cell viability, while siRNA targeted to SSeCKS in astrocytes inhibited oligodendrocyte cell death. The results from antibody neutralization and NO inhibition suggested that the oligodendrocyte apoptosis may be due to the production of astrocyte-derived proinflammatory factors (TNF-alpha and NO). These findings revealed that there was a pathogenic interaction between SSeCKS expression in astrocytes and oligodendrocyte apoptosis. Understanding the mechanism of SSeCKS in the pathogenesis of EAE may contribute to the development of new therapeutic strategies against EAE and MS.

Cyclin D3/CDK11(p58) complex involved in Schwann cells proliferation repression caused by lipopolysaccharide.

Schwann cells proliferation is the main characterize of kinds PNS inflammation diseases. It has been well documented that cyclin D3 /CDK11(p58) complex inhibits cell function through multiple mechanisms, but the mechanism of cyclin D3/CDK11(p58) complex exerts its repressive role in the Schwann cells proliferation remains to be identified. In the present investigation, we demonstrated that the expression of CDK11(p58) were upregulated in the inflammation caused by LPS, a main part of bacteria. Cyclin D3 and the 58-kDa isoform of cyclin-dependent kinase 11 (CDK11(p58)) interacted with each other mainly in nuclear region, repressed Schwann cells proliferation and induced cell apoptosis. Overexpression of CDK11(p58) expression might enhance this process, while silence of cyclin D3 reverting it. This work demonstrates for the first time the role of cyclin D3/CDK11(p58) complex in repressing the Schwann cells proliferation and inducing its apoptosis.

Peroxisome proliferator-activated receptor-gamma agonists suppress iNOS expression induced by LPS in rat primary Schwann cells.

In bacterial-induced peripheral nervous system (PNS) inflammation, Schwann cells (SCs) are activated, producing inducible nitric oxide synthase (iNOS), contributed to the pathogenesis of demyelinating disease, such as multiple sclerosis. Peroxisome proliferator-activated receptor-gamma (PPAR-gamma) has been shown to play a protective role in cellular inflammatory responses. Here we showed that LPS-induced iNOS biosynthesis was in a concentration and time-dependent manner. In LPS-treated primary SCs, retreatment with PPAR-gamma agonist remitted the increase of iNOS, p38 phosphorylation and TLR4, MyD88, augmented the expression of PPAR-gamma and localization in nuclear. Coadministration of GW 9662 reversed the effect of PPAR-gamma agonists. These results suggest that PPAR-gamma agonists, 15d-PGJ(2) and pioglitazone, had the anti-inflammatory effects.

Lipopolysaccharide induced upregulation of beta-1,4-galactosyltransferase-I in Schwann cell.

beta4 Galactosylation of glycoproteins is one of the most important post-translational modifications. Recent studies have demonstrated that aberrant galactosylation associates with some inflammation diseases. beta-1,4-galactosyltransferase-I (beta-1,4-GalT-I), which transfers galactose to the terminal N-acetylglucosamine of N- and O-linked glycans in a beta-1,4- linkage, considered to be the major galactosyltransferse among the seven members of the subfamily responsible for beta4 galactosylation. In the present study, we investigated the expression of beta-1,4-GalT-I in Schwann cells under Lipopolysaccharide (LPS) treatment. RT-PCR revealed that the beta-1,4-GalT-I mRNA was significant increased as early as 2 h after LPS stimulation. Immunofluorescence showed that beta-1,4-GalT-I was located in Golgi apparatus and membrane of Schwann cells. With the 1 microg/ml LPS treatment, expression levels of beta-1,4-GalT-I was much higher compared with control group. In addition, lectin blot indicated that the beta4 galactosylation of glycoproteins such as integrin alpha5 was enhanced, which may due to the induced beta-1,4-GalT-I expression. These results suggested that beta-1,4-GalT-I may play an important role in adhesion and migration of Schwann cells during inflammation.