Cyclic Self-Training with Proposal Weight Modulation for Cross-Supervised Object Detection.

Weakly-supervised object detection (WSOD), which requires only image-level annotations for training detectors, has gained enormous attention. Despite recent rapid advance in WSOD, there remains a large performance gap compared with fully-supervised object detection. To narrow the performance gap, we study cross-supervised object detection (CSOD), where existing classes (base classes) have instance-level annotations while newly added classes (novel classes) only need image-level annotations. For improving localization accuracy, we propose a Cyclic Self-Training (CST) method to introduce instance-level supervision into a commonly used WSOD method, online instance classifier refinement (OICR). Our proposed CST consists of forward pseudo labeling and backward pseudo labeling. Specifically, OICR exploits the forward pseudo labeling to generate pseudo ground-truth bounding-boxes for all classes, thus enabling instance classifier training. Then, the backward pseudo labeling is designed to generate pseudo ground-truth bounding-boxes of higher quality for novel classes by fusing the predictions of the instance classifiers. As a result, both novel and base classes will have bounding-box annotations for training, alleviating the supervision inconsistency between base and novel classes. In the forward pseudo labeling, the generated pseudo ground-truths may be misaligned with objects and thus introduce poor-quality examples for training the ICs. To reduce the impacts of these poor-quality training examples, we propose a Proposal Weight Modulation (PWM) module learned in a class-agnostic and contrastive manner by exploiting bounding-box annotations of base classes. Experiments on PASCAL VOC and MS COCO datasets demonstrate the superiority of our proposed method.

A Phenomenological Study of Nurses' Experiences in a Pediatric Fever Clinic During the COVID-19 Epidemic.

BACKGROUND AND PURPOSE: In the initial COVID-19 outbreak, nursing staff reportedly experienced high levels of psychological stress. The purpose of this study was to explore the real experience of the first cohort of pediatric fever clinic nurses during the COVID-19 epidemic. METHODS: Semi-structured interviews were conducted with eight nurses who worked in a fever clinic at a children's hospital in China. The interviews were conducted by an experienced and trained interviewer. Qualitative content analysis was used to describe the experiences of the nurses. RESULTS: Three themes were distilled from the interviews: 1) complex psychological experiences including positive experiences (increased sense of responsibility and honor, gaining the respect and recognition of parents, having a sense of achievement in personal growth) and negative experiences (panic and compulsion, guilt towards their family, antipathy, and dissatisfaction); 2) extreme physical discomfort; and 3) a lack of relevant knowledge. IMPLICATIONS FOR PRACTICE: The nurses in the pediatric fever clinic experienced various psychological impacts and physiological discomfort. Nursing managers should improve the management of hospital emergency nursing, strengthen the psychological guidance and logistics support of frontline nurses, and provide nurses with the relevant knowledge and skills training. These improvements would support frontline nurses in their work to provide effective patient treatment during the COVID-19 epidemic.

Enhanced microbial nitrification-denitrification processes in a subtropical metropolitan river network.

Urban rivers are hotspots of regional nitrogen (N) pollution and N transformations. Previous studies have reported that the microbial community of urban rivers was different from that of natural rivers. However, how microbial community affects N transformations in the urban rivers is still unclear. In this study, we employed N nutrients-related isotope technology (includes natural-abundance isotopes survey and isotope-labeling method) and bioinformatics methods (includes 16S rRNA high-throughput sequencing and quantitative PCR analysis) to investigate the major N transformations, microbial communities as well as functional gene abundances in a metropolitan river network. Our results suggested that the bacterial community structure in the highly urbanized rivers was characterized by higher richness, less complexity and increased abundances of nitrification and denitrifying bacterium compared to those in the suburban rivers. These differences were mainly caused by high sewage discharge and N loadings. In addition, the abundances of nitrifier gene (amoA) and denitrifier genes (nirK and nirS) were significantly higher in the highly urbanized rivers (2.36 × 103, 7.43 × 107 and 2.28 × 107 copies·mL-1) than that in the suburban rivers (0.43 × 103, 2.18 × 107 and 0.99 × 107 copies·mL-1). These changes in microbes have accelerated nitrification-denitrification processes in the highly urbanized rivers as compared to those in the suburban rivers, which was evidenced by environmental isotopes and the rates of nitrification (10.52 vs. 0.03 nmol·L-1·h-1) and denitrification (83.31 vs. 22.49 nmol·g-1·h-1). Overall, this study concluded that the excess exogenous N has significantly shaped the specific aquatic bacterial communities, which had a potential for enhancing nitrification-denitrification processes in the highly urbanized river network. This study provides a further understanding of microbial N cycling in urban river ecosystems and expands the combined application of isotopic technology and bioinformatics methods in studying biogeochemical cycling.

SIRT4 functions as a tumor suppressor during prostate cancer by inducing apoptosis and inhibiting glutamine metabolism.

Localized in the mitochondria, SIRT4 is a nicotinamide adenine dinucleotide (NAD +) -dependent adenosine diphosphate (ADP) -ribosyltransferase and is one of the least characterized members of the sirtuin family. Although it is well known that it shows deacetylase activity for energy metabolism, little is understood about its function in tumorigenesis. Recent research suggests that SIRT4 may work as both a tumor suppressor gene and an oncogene. However, the clinical significance of SIRT4 in prostate cancer remains unknown. In this study, we evaluated SIRT4 protein levels in cancerous prostate tissue and corresponding non-tumor prostate tissue via immunohistochemical staining on a tissue microarray including tissues from 89 prostate cancer patients. The association between SIRT4 expression and Gleason score was also determined. Further, shSIRT4 or stable prostate cancer cell lines (22RV1) overexpressing SIRT4 were constructed via lentiviral infection. Using Cell-Counting Kit-8 (CCK-8) assay, wound healing assay, migration, and invasion and apoptosis assays, the effects of SIRT4 on the migration, invasion ability, and proliferation of prostate cancer cells were investigated. We also determined the effect of SIRT4 on glutamine metabolism in 22RV1 cells. We found the protein levels of SIRT4 in prostate cancer tissues were significantly lower than those in their non-neoplastic tissue counterparts (P < 0.01); a lower SIRT4 level was also significantly associated with a higher Gleason score (P < 0.01). SIRT4 suppressed the migration, invasion capabilities, and proliferation of prostate cancer cells and induced cellular apoptosis. Furthermore, the invasion and migration of 22RV1 cells were mechanistically inhibited by SIRT4 via glutamine metabolism inhibition. In conclusion, the present study's findings showed that SIRT4 protein levels are significantly associated with the Gleason score in patients with prostate cancer, and SIRT4 exerts a tumor-suppressive effect on prostate cancer cells by inhibiting glutamine metabolism. Thus, SIRT4 may serve as a potential novel therapeutic target for prostate cancer.

Weakly Supervised RGB-D Salient Object Detection With Prediction Consistency Training and Active Scribble Boosting.

RGB-D salient object detection (SOD) has attracted increasingly more attention as it shows more robust results in complex scenes compared with RGB SOD. However, state-of-the-art RGB-D SOD approaches heavily rely on a large amount of pixel-wise annotated data for training. Such densely labeled annotations are often labor-intensive and costly. To reduce the annotation burden, we investigate RGB-D SOD from a weakly supervised perspective. More specifically, we use annotator-friendly scribble annotations as supervision signals for model training. Since scribble annotations are much sparser compared to ground-truth masks, some critical object structure information might be neglected. To preserve such structure information, we explicitly exploit the complementary edge information from two modalities (i.e., RGB and depth). Specifically, we leverage the dual-modal edge guidance and introduce a new network architecture with a dual-edge detection module and a modality-aware feature fusion module. In order to use the useful information of unlabeled pixels, we introduce a prediction consistency training scheme by comparing the predictions of two networks optimized by different strategies. Moreover, we develop an active scribble boosting strategy to provide extra supervision signals with negligible annotation cost, leading to significant SOD performance improvement. Extensive experiments on seven benchmarks validate the superiority of our proposed method. Remarkably, the proposed method with scribble annotations achieves competitive performance in comparison to fully supervised state-of-the-art methods.

Self-Correction for Human Parsing.

Labeling pixel-level masks for fine-grained semantic segmentation tasks, e.g., human parsing, remains a challenging task. The ambiguous boundary between different semantic parts and those categories with similar appearances are usually confusing for annotators, leading to incorrect labels in ground-truth masks. These label noises will inevitably harm the training process and decrease the performance of the learned models. To tackle this issue, we introduce a noise-tolerant method in this work, called Self-Correction for Human Parsing (SCHP), to progressively promote the reliability of the supervised labels as well as the learned models. In particular, starting from a model trained with inaccurate annotations as initialization, we design a cyclically learning scheduler to infer more reliable pseudo masks by iteratively aggregating the current learned model with the former sub-optimal one in an online manner. Besides, those correspondingly corrected labels can in turn to further boost the model performance. In this way, the models and the labels will reciprocally become more robust and accurate during the self-correction learning cycles. Our SCHP is model-agnostic and can be applied to any human parsing models for further enhancing their performance. Extensive experiments on four human parsing models, including Deeplab V3+, CE2P, OCR and CE2P+, well demonstrate the effectiveness of the proposed SCHP. We achieve the new state-of-the-art results on 6 benchmarks, including LIP, Pascal-Person-Part and ATR for single human parsing, CIHP and MHP for multi-person human parsing and VIP for video human parsing tasks. In addition, benefiting the superiority of SCHP, we achieved the 1st place on all the three human parsing tracks in the 3rd Look Into Person Challenge. The code is available at https://github.com/PeikeLi/Self-Correction-Human-Parsing.

Reinforcement Learning With Multiple Relational Attention for Solving Vehicle Routing Problems.

In this article, we study the reinforcement learning (RL) for vehicle routing problems (VRPs). Recent works have shown that attention-based RL models outperform recurrent neural network-based methods on these problems in terms of both effectiveness and efficiency. However, existing RL models simply aggregate node embeddings to generate the context embedding without taking into account the dynamic network structures, making them incapable of modeling the state transition and action selection dynamics. In this work, we develop a new attention-based RL model that provides enhanced node embeddings via batch normalization reordering and gate aggregation, as well as dynamic-aware context embedding through an attentive aggregation module on multiple relational structures. We conduct experiments on five types of VRPs: 1) travelling salesman problem (TSP); 2) capacitated VRP (CVRP); 3) split delivery VRP (SDVRP); 4) orienteering problem (OP); and 5) prize collecting TSP (PCTSP). The results show that our model not only outperforms the learning-based baselines but also solves the problems much faster than the traditional baselines. In addition, our model shows improved generalizability when being evaluated in large-scale problems, as well as problems with different data distributions.

Brusatol Inhibits Tumor Growth and Increases the Efficacy of Cabergoline against Pituitary Adenomas.

Cabergoline (CAB) is the first choice for treatment of prolactinoma and the most common subtype of pituitary adenoma. However, drug resistance and lack of effectiveness in other pituitary tumor types remain clinical challenges to this treatment. Brusatol (BT) is known to inhibit cell growth and promote apoptosis in a variety of cancer cells. In our present studies, we investigate the effects of BT on pituitary tumor cell proliferation in vitro and in vivo. BT treatment resulted in an increase in Annexin V-expressing cells and promoted the expression of apoptosis-related proteins in rat and human pituitary tumor cells. Investigation of the mechanism underlying this effect revealed that BT increased the production of reactive oxygen species (ROS) and inhibited the phosphorylation of 4EBP1 and S6K1. Furthermore, treatment with a combination of BT and CAB resulted in greater antitumor effects than either treatment alone in nude mice and pituitary tumor cells. Collectively, our results suggest that the BT-induced ROS accumulation and inhibition of mTORC1 signaling pathway leads to inhibition of tumor growth. Combined use of CAB and BT may increase the clinical effectiveness of treatment for human pituitary adenomas.

Investigation and Analysis of Perceived Stress Levels in Pediatric Nurses During the 2019-20 SARS-CoV-2 Outbreak.

This study investigated perceived stress levels of pediatric nurses during the 2019-20 severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreak. From February 9 to 13, 2020, 250 pediatric nurses were selected from 4 grade III children's hospitals and 5 grade III general hospitals in Jiangsu Province, China. A general information questionnaire and a stress perception scale were used to investigate and analyze the influencing factors of perceived stress among participants. Perceived stress was at the intermediate level or above for 54.1% (133) of the sample. Multiple linear regression analysis showed that hospital department, living styles, physical conditions, and the proportion of negative (eg, increased deaths and infections) information received about the epidemic (eg, via media and/or face-to-face) influenced perceived stress. Thus, the overall level of perceived stress among clinical pediatric nurses was medium. Nursing managers should pay more attention to and conduct timely assessment interventions for staff to reduce stress levels and promote physical and mental health.

Pyramidal Multiple Instance Detection Network With Mask Guided Self-Correction for Weakly Supervised Object Detection.

Weakly supervised object detection has attracted more and more attention as it only needs image-level annotations for training object detectors. A popular solution to this task is to train a multiple instance detection network (MIDN) which integrates multiple instance learning into a deep convolutional neural network. One major issue of the MIDN is that it is prone to be stuck at local discriminative regions. To address this local optimum issue, we propose a pyramidal MIDN (P-MIDN) comprised of a sequence of multiple MIDNs. In particular, one MIDN performs proposal removal for its subsequent MIDN to reduce the exposure of local discriminative proposal regions to the latter during training. In this manner, it allows our MIDNs to focus on proposals which cover objects more completely. Furthermore, we integrate the P-MIDN into an online instance classifier refinement (OICR) framework. Combined with the P-MIDN, a mask guided self-correction (MGSC) method is proposed to generate high-quality pseudo ground-truths for training the OICR. Experimental results on PASCAL VOC 2007, PASCAL VOC 2010, PASCAL VOC 2012, ILSVRC 2013 DET and MS-COCO benchmarks demonstrate that our approach achieves state-of-the-art performance.

Pre-treatment Glasgow prognostic score and modified Glasgow prognostic score may be potential prognostic biomarkers in urological cancers: a systematic review and meta-analysis.

BACKGROUND: The prognostic role of Glasgow prognostic score (GPS) or modified GPS (mGPS) in various cancers has been investigated. However, no unified conclusion could be drawn in urological cancers. So, we aimed to explore the potential role of GPS/mGPS in urological cancers. METHODS: Related studies were searched from PubMed, Web of Science and Embase up to May 30th, 2019 comprehensively. Their associations were assessed by the pooled hazard ratios (HRs) with its 95% confidence intervals (CIs). RESULTS: A total of 20 related studies were enrolled in this meta-analysis. The outcomes revealed that a relatively lower level of pre-treatment GPS/mGPS was associated with better overall survival (OS), cancer specific survival (CSS)/disease-specific survival (DSS) and disease-free survival (DFS)/progress-free survival (PFS)/recurrence-free survival (RFS) (pooled HR =2.70; 95% CI, 1.81-4.01; pooled HR =2.90; 95% CI, 2.00-4.22; pooled HR =2.43; 95% CI, 1.62-3.66, respectively). Subgroup analysis by cancer type for OS indicated that GPS/mGPS could also be a predictor no matter in renal cell cancer (RCC) or bladder cancer (BC) (pooled HR =3.60; 95% CI, 2.07-6.28 and pooled HR =2.71; 95% CI, 1.08-6.82). Similar results could be found in CSS/DSS (RCC: HR =4.12; 95% CI, 2.69-6.30) and in DFS/ PFS/RFS (RCC: HR =2.66; 95% CI, 1.82-3.90 and BC: HR =1.52; 95% CI, 1.23-1.88). As for the treatment subgroup, pre-treatment GPS/mGPS played an independent role in OS for patients no matter in which treatment type (Surgery: pooled HR =2.16; 95% CI, 1.43-3.26; Chemotherapy: pooled HR =4.41; 95% CI, 2.27-8.58); the same in CSS/DSS (Surgery: pooled HR =3.28; 95% CI, 1.73-6.20; Immunotherapy: pooled HR =2.72; 95% CI, 1.87-3.96) and DFS/RFS/PFS (Surgery: pooled HR =2.54; 95% CI, 1.65-3.92). Lastly, both GPS and mGPS played prognostic role in OS, CSS/DSS or DFE/RFS/PFS (OS: GPS: pooled HR =2.12; 95% CI, 1.04-4.32; mGPS: pooled HR =3.12; 95% CI, 1.87-5.20; CSS/DSS: GPS: pooled HR =2.87; 95% CI, 2.11-3.91; mGPS: pooled HR =3.00; 95% CI, 1.60-5.63; DFS/RFS/PFS: GPS: pooled HR =3.61; 95% CI, 1.43-9.07; mGPS: pooled HR =1.99; 95% CI, 1.32-2.99). CONCLUSIONS: This study shed light on that GPS/mGPS might be an independent prognostic factor in urological cancers, indicating that a lower level of pre-treatment GPS/mGPS was closely related to better survival outcomes.