Semantic Image Inpainting with Multi-Stage Feature Reasoning Generative Adversarial Network.

Most existing image inpainting methods have achieved remarkable progress in small image defects. However, repairing large missing regions with insufficient context information is still an intractable problem. In this paper, a Multi-stage Feature Reasoning Generative Adversarial Network to gradually restore irregular holes is proposed. Specifically, dynamic partial convolution is used to adaptively adjust the restoration proportion during inpainting progress, which strengthens the correlation between valid and invalid pixels. In the decoding phase, the statistical natures of features in the masked areas differentiate from those of unmasked areas. To this end, a novel decoder is designed which not only dynamically assigns a scaling factor and bias on per feature point basis using point-wise normalization, but also utilizes skip connections to solve the problem of information loss between the codec network layers. Moreover, in order to eliminate gradient vanishing and increase the reasoning times, a hybrid weighted merging method consisting of a hard weight map and a soft weight map is proposed to ensemble the feature maps generated during the whole reconstruction process. Experiments on CelebA, Places2, and Paris StreetView show that the proposed model generates results with a PSNR improvement of 0.3 dB to 1.2 dB compared to other methods.

Domain Adaptive Box-Supervised Instance Segmentation Network for Mitosis Detection.

The number of mitotic cells present in histopathological slides is an important predictor of tumor proliferation in the diagnosis of breast cancer. However, the current approaches can hardly perform precise pixel-level prediction for mitosis datasets with only weak labels (i.e., only provide the centroid location of mitotic cells), and take no account of the large domain gap across histopathological slides from different pathology laboratories. In this work, we propose a Domain adaptive Box-supervised Instance segmentation Network (DBIN) to address the above issues. In DBIN, we propose a high-performance Box-supervised Instance-Aware (BIA) head with the core idea of redesigning three box-supervised mask loss terms. Furthermore, we add a Pseudo-Mask-supervised Semantic (PMS) head for enriching characteristics extracted from underlying feature maps. Besides, we align the pixel-level feature distributions between source and target domains by a Cross-Domain Adaptive Module (CDAM), so as to adapt the detector learned from one lab can work well on unlabeled data from another lab. The proposed method achieves state-of-the-art performance across four mainstream datasets. A series of analysis and experiments show that our proposed BIA and PMS head can accomplish mitosis pixel-wise localization under weak supervision, and we can boost the generalization ability of our model by CDAM.

Gold nano-particles (AuNPs) carrying miR-326 targets PDK1/AKT/c-myc axis in hepatocellular carcinoma.

Abnormal expression of microRNAs (miRNAs) contributes to tumour growth and invasion. MiR-326 expression often down-regulates in several kinds of cancer and low expression of miR-326 is linked with poor prognosis in cancer patients. In the present study, we aimed to explore the modulatory mechanism of miR-326 in hepatocellular carcinoma (HCC). miR-326 expression was significantly decreased in HCC cell lines and tissues. miR-326 decreased HCC cell growth by affecting cell-cycle progression and by promoting apoptosis. In addition, miR-326 inhibited HCC cell invasion by decreasing the EMT phenotype. We found that miR-326 functioned as a tumour suppressor by repressing its down-stream target PDK1. C-myc contributed to miR-326 down-regulation through binding at its promoter and inhibited its expression. Based on these results, we conducted a therapeutic experiment by using gold nano-particles (AuNPs) carrying miR-326. Restoration of miR-326 reduced tumour growth in vivo. Our findings suggest that miR-326 may be a candidate prognostic biomarker and a target for new therapies in HCC patients.

LINC01287 regulates tumorigenesis and invasion via miR-298/MYB in hepatocellular carcinoma.

Recently, it was reported that long non-coding RNAs (lncRNAs) participated in promoting hepatocellular carcinoma (HCC) initiation and progression. Herein, we reported that the expression level of LINC01287 was elevated in HCC cell lines and tissues. LINC01287 down-regulation inhibited HCC cells growth and invasion both in vitro and in vivo. LINC01287 exerted as a ceRNA and negatively regulated miR-298 expression. MYB was identified as a downstream target of miR-298. The miR-298/MYB axis mediated LINC01287's effect on HCC. To the best of our knowledge, our findings provided the first evidence that LINC01287 functioned as an oncogene in HCC. LINC01287 may be a candidate prognostic biomarker and a target for new therapies in HCC patients.

LINC01287/miR-298/STAT3 feedback loop regulates growth and the epithelial-to-mesenchymal transition phenotype in hepatocellular carcinoma cells.

BACKGROUND: The long non-coding RNAs (lncRNAs) have participated in the promotion of hepatocellular carcinoma (HCC) initiation and progression. Nevertheless, the biological role and underlying mechanism of LINC01287 in HCC has never been reported. METHODS: The TGCA database was used to explore the abnormal expression of lncRNAs in HCC. Real-time PCR and in situ hybridization assays were used to examine the expression of LINC01287 in HCC tissues. The clinicopathological characteristics of HCC patients in relation to LINC01287 expression were then analyzed. Infection of cells with the si-LINC01287 lentiviral vector was performed to down-regulate LINC01287 expression in HCC cells. MTT and colony formation assays were performed to examine cell growth ability, and FACS analysis was performed to examine the cell cycle and apoptosis. A Boyden assay was used to examine HCC cell invasion ability, and RNA immunoprecipitation tested the interaction between LINC01287 and miR-298. A luciferase reporter assay was used to examine whether STAT3 was a direct target of miR-298, and chromatin immunoprecipitation (ChIP) was used to examine the potential binding of c-jun to the miR-298 promoter. RESULTS: We revealed that the expression of LINC01287 was increased in HCC cell lines, as well as tissues. Knockdown of LINC01287 decreased HCC cell growth and invasion both in vitro and in vivo. LINC01287 can negatively regulate miR-298 expression by acting as a ceRNA. miR-298 directly targeted STAT3 and inhibited its expression. LINC01287 exerted its function via the miR-298/STAT3 axis in HCC. Interestingly, STAT3 elevated LINC01287 expression via c-jun, which bound to the LINC01287 promoter. A feedback loop was also discovered between LINC01287 and the miR-298/STAT3 axis. CONCLUSIONS: Our data indicated that LINC01287 played an oncogenic role in HCC growth and metastasis and that this lncRNA might serve as a novel molecular target for the treatment of HCC.