A lightweight network for traffic sign recognition based on multi-scale feature and attention mechanism.

Traffic sign recognition is an important part of intelligent transportation system. It uses computer vision and traffic sign recognition technology to detect and recognize traffic signs on the road automatically. In this paper, we propose a lightweight model for traffic sign recognition based on convolutional neural networks called ConvNeSe. Firstly, the feature extraction module of the model is constructed using the Depthwise Separable Convolution and Inverted Residuals structures. The model extracts multi-scale features with strong representation ability by optimizing the structure of convolutional neural networks and fusing of features. Then, the model introduces Squeeze and Excitation Block (SE Block) to improve the attention to important features, which can capture key information of traffic sign images. Finally, the accuracy of the model in the German Traffic Sign Recognition Benchmark Database (GTSRB) is 99.85%. At the same time, the model has good robustness according to the results of ablation experiments.

Mining and validation of prognosis of various visceral metastasis in renal cell carcinoma: a study based on SEER database.

Our study was aimed to analyze a substantial of renal cell carcinoma (RCC) patients, research the high-risk factors and prognostic factors of metastasis, and thoroughly examine the effects of primary site surgery, lymph node dissection (LND), and chemotherapy on the prognosis of different visceral metastases. The baseline characteristics were characterized, and logistic regression was used to predict the risk factors for metastasis. Prognostic factors of metastatic RCC were assessed using batch univariate and multivariate Cox regression, with adjustments made through PSM. Next, the Kaplan-Meier method was employed to assess OS and create the survival curve. Logistic regression identified risk factors for metastasis: male gender [OR, 1.223; P < 0.001], Hist clear (OR, 9.37; P < 0.001), Hist papillary (OR, 2.49; P < 0.001), and TTX (OR, 23.33; P < 0.001). We found several independent prognostic variables: among which chemotherapy (HR, 0.64), local LND (HR, 0.67), and primary site surgery (HR, 0.97) were associated with better OS. Further study results demonstrated that all kinds of visceral metastasis except for liver metastasis in the operation group had substantially better prognoses than those in the non-operation group (P < 0.05). Regional LND had no discernible impact on survival. Patients with liver, lung, and distant lymph node (LN) metastasis benefited from chemotherapy (P < 0.05), but the bone and brain metastasis did not significantly benefit from treatment (P > 0.05). We recommend primary surgery for different types of visceral metastases except liver metastasis. Routine regional LND is not recommended. Chemotherapy should be considered for patients with lung, distant LN, and liver metastases, but not for those with bone and brain metastases.

The underlying relationship between exercise and the prevalence of periodontitis: a systematic review and meta-analysis.

BACKGROUND: Although exercise has been proposed as an effective intervention for various lifestyle-related diseases and pathological scenarios, few researches assessed the impact of taking exercise on the prevalence of periodontitis. PURPOSE: This study aimed to perform a comprehensive literature review and meta-analysis from both observational and intervention studies to explore the relationship between exercise and periodontitis and to provide references for future intervention programs aimed at preventing periodontitis. METHOD: A systematic literature search was conducted in PubMed/MEDLINE, Web of Science, Cochrane Library, and Scopus for peer-reviewed studies published in English From January 1993 to January 2023 according to the PRISMA guidelines. Articles were selected if subjects were human and studies evaluating the association between exercise and periodontitis. RESULTS: 4098 references were retrieved. After screening the results, 30 studies were selected. Of these, 20 studies indicated an inverse association between exercise and periodontitis, while the remaining 10 did not reach conclusive findings. The meta-analysis demonstrated a risk ratio of 0.84 (95%CI: 0.77, 0.91) between the active group and the inactive group (P < 0.01), which indicated an inverse relationship between exercise and periodontitis. CONCLUSION: Overall, the increase in exercise presents an inverse association with the presence and severity of periodontitis. Accordingly, taking exercise might be a potential approach that contributes to improvements in periodontitis.

S100 proteins in head and neck squamous cell carcinoma (Review).

The most common tumor affecting the head and neck is head and neck squamous cell carcinoma (HNSCC). The characteristics of HNSCC include a rapid onset, a lack of early diagnosis, drug resistance, relapse and systemic adverse effects, leading to inadequate prevention, diagnosis and treatment. Notably, previous research suggests that there is an association between S100 proteins and HNSCC. S100A8, S100A9 and S100A14 interfere with tumor cell proliferation by blocking the cell cycle. The present review discusses this association. S100A4 enhances cancer stem cell properties, and interacts with actin and tropomyosin to promote tumor cell migration. S100A1, S100A8, S100A9, S100A10, S100A14 and S100P are involved in the initiation and progression of HNSCC via Hippo, nuclear factor κB, phosphatidylinositol kinase/protein kinase B/mammalian target of rapamycin and other signaling pathways. In addition, certain long non-coding RNAs and microRNAs are involved in regulating the expression of S100 proteins in HNSCC. Reducing the expression of certain members of the S100 protein family may enhance the chemosensitivity of HNSCC. Collectively, it is suggested that S100 proteins may function as markers and targets for the prevention, diagnosis and treatment of HNSCC.

Human CPTP promotes growth and metastasis via sphingolipid metabolite ceramide and PI4KA/AKT signaling in pancreatic cancer cells.

Pancreatic cancer (PC) is a devastating solid malignancy with a dismal prognosis. The treatment of metastatic PC is a current challenge for medical oncologists due to a lack of early detection, drug resistance, and relapse. Therefore, potential biomarkers and effective therapeutic targets for PC are urgently required. Ceramide-1-phosphate transfer protein (CPTP) is a member of the glycolipid transfer protein family, which is associated with autophagy and inflammation regulation. The roles and mechanisms of CPTP in PC have not been clarified. In this study, by RT-qPCR and immunohistochemistry analysis, we found that CPTP is highly expressed in PC and is associated with a poor prognosis in PC patients. By using cell counting kit-8, colony formation, transwell and matrigel assays in vitro, as well as xenograft model assays in vivo, we further proved that CPTP enhanced PC cells growth and metastasis. In PC cells, human CPTP promotes growth and metastasis via sphingolipid metabolite ceramide and PI4KA/AKT signaling. Sp (specific protein)-1 and Sp3 transcription factors also act as upstream positive regulators of CPTP expression in PC cells. Collectively, these findings suggested that CPTP may function as a pro-tumorigenic gene in PC cells and could be a promising therapeutic target in PC.

Relational Reasoning Over Spatial-Temporal Graphs for Video Summarization.

In this paper, we propose a dynamic graph modeling approach to learn spatial-temporal representations for video summarization. Most existing video summarization methods extract image-level features with ImageNet pre-trained deep models. Differently, our method exploits object-level and relation-level information to capture spatial-temporal dependencies. Specifically, our method builds spatial graphs on the detected object proposals. Then, we construct a temporal graph by using the aggregated representations of spatial graphs. Afterward, we perform relational reasoning over spatial and temporal graphs with graph convolutional networks and extract spatial-temporal representations for importance score prediction and key shot selection. To eliminate relation clutters caused by densely connected nodes, we further design a self-attention edge pooling module, which disregards meaningless relations of graphs. We conduct extensive experiments on two popular benchmarks, including the SumMe and TVSum datasets. Experimental results demonstrate that the proposed method achieves superior performance against state-of-the-art video summarization methods.