tRFs as Potential Exosome tRNA-Derived Fragment Biomarkers for Gastric Carcinoma.

BACKGROUND: Gastric Carcinoma (GC) is one of the common diseases induced by the interaction of genes and environment. Exosomes are potential markers for several health problems, which contain lipids, proteins, long non-coding RNAs, microRNAs (miRNAs), and tRNA-derived fragments (tRFs). The roles of mRNAs and miRNAs in GC have been studied comprehensively; however, little research was focused on the function of plasma exosomal tRFs. METHODS: We collected plasma samples from fifty healthy controls and fifty GC patients, and all exosomes were isolated with a combined centrifugation and characterized by electron microscopy, western blot, and flow cytometry. The small RNA sequence was performed to detect the plasma exosomal tRFs, and tRFs markers were validated by real-time quantitative PCR. Three exosomal diagnostic tRFs were confirmed by receiver operating characteristic analyses. RESULTS: In this study, we found higher plasma exosomal tRF-25, tRF-38, tRF-18 expression in GC than in controls. Plasma exosomal tRF-25, tRF-38, and tRF-18 showed better accuracy for GC diagnosis. CONCLUSIONS: Our results suggest that plasma exosomal tRF-25, tRF-38, and tRF-18 were biomarkers for GC detection; tRF-25, tRF-38 and tRF-18 might be predictive of GC prognosis.

VNVC: A Versatile Neural Video Coding Framework for Efficient Human-Machine Vision.

Almost all digital videos are coded into compact representations before being transmitted. Such compact representations need to be decoded back to pixels before being displayed to humans and - as usual - before being enhanced/analyzed by machine vision algorithms. Intuitively, it is more efficient to enhance/analyze the coded representations directly without decoding them into pixels. Therefore, we propose a versatile neural video coding (VNVC) framework, which targets learning compact representations to support both reconstruction and direct enhancement/analysis, thereby being versatile for both human and machine vision. Our VNVC framework has a feature-based compression loop. In the loop, one frame is encoded into compact representations and decoded to an intermediate feature that is obtained before performing reconstruction. The intermediate feature can be used as reference in motion compensation and motion estimation through feature-based temporal context mining and cross-domain motion encoder-decoder to compress the following frames. The intermediate feature is directly fed into video reconstruction, video enhancement, and video analysis networks to evaluate its effectiveness. The evaluation shows that our framework with the intermediate feature achieves high compression efficiency for video reconstruction and satisfactory task performances with lower complexities.

Self-Supervised Representation Learning with Spatial-Temporal Consistency for Sign Language Recognition.

Recently, there have been efforts to improve the performance in sign language recognition by designing self-supervised learning methods. However, these methods capture limited information from sign pose data in a frame-wise learning manner, leading to sub-optimal solutions. To this end, we propose a simple yet effective self-supervised contrastive learning framework to excavate rich context via spatial-temporal consistency from two distinct perspectives and learn instance discriminative representation for sign language recognition. On one hand, since the semantics of sign language are expressed by the cooperation of fine-grained hands and coarse-grained trunks, we utilize both granularity information and encode them into latent spaces. The consistency between hand and trunk features is constrained to encourage learning consistent representation of instance samples. On the other hand, inspired by the complementary property of motion and joint modalities, we first introduce first-order motion information into sign language modeling. Additionally, we further bridge the interaction between the embedding spaces of both modalities, facilitating bidirectional knowledge transfer to enhance sign language representation. Our method is evaluated with extensive experiments on four public benchmarks, and achieves new state-of-the-art performance with a notable margin. The source code are publicly available at https://github.com/sakura/Code.

A Robust Framework for One-Shot Key Information Extraction via Deep Partial Graph Matching.

Text field labelling plays a key role in Key Information Extraction (KIE) from structured document images. However, existing methods ignore the field drift and outlier problems, which limit their performance and make them less robust. This paper casts the text field labelling problem into a partial graph matching problem and proposes an end-to-end trainable framework called Deep Partial Graph Matching (dPGM) for the one-shot KIE task. It represents each document as a graph and estimates the correspondence between text fields from different documents by maximizing the graph similarity of different documents. Our framework obtains a strict one-to-one correspondence by adopting a combinatorial solver module with an extra one-to-(at most)-one mapping constraint to do the exact graph matching, which leads to the robustness of the field drift problem and the outlier problem. Finally, a large one-shot KIE dataset named DKIE is collected and annotated to promote research of the KIE task. This dataset will be released to the research and industry communities. Extensive experiments on both the public and our new DKIE datasets show that our method can achieve state-of-the-art performance and is more robust than existing methods.

Gastric epithelial neoplasm of fundic-gland mucosa lineage: representative of the low atypia differentiated gastric tumor and Ki67 may help in their identification.

Background: Gastric epithelial neoplasm of the fundic-gland mucosa lineages (GEN-FGMLs) are rare forms of gastric tumors that encompass oxyntic gland adenoma (OGA), gastric adenocarcinoma of the fundic-gland type (GA-FG), and gastric adenocarcinoma of the fundic-gland mucosa type (GA-FGM). There is no consensus on the cause, classification, and clinicopathological features of GEN-FGMLs, and misdiagnosis is common because of similarities in symptoms. Methods: 37 cases diagnosed with GEN-FGMLs were included in this study. H&E-stained slides were reviewed and clinicopathological parameters were recorded. Immunohistochemical staining was conducted for MUC2, MUC5AC, MUC6, CD10, CD56, synaptophysin, chromograninA, p53, Ki67, pepsinogen-I, H+/K+-ATPase and Desmin. Results: The patients' ages ranged from 42 to 79 years, with a median age of 60. 17 were male and 20 were female. Morphologically, 19 OGAs, 16 GA-FGs, and two GA-FGMs were identified. Histopathological similarities exist between OGA, GA-FG, and GA-FGM. The tumors demonstrated well-formed glands, expanding with dense growth patterns comprising pale, blue-grey columnar cells with mild nuclear atypia. These cells resembled fundic gland cells. None of the OGA invaded the submucosal layer. The normal gastric pit epithelium covered the entire surface of the OGA and GA-FG, but the dysplasia pit epithelium covered the GA-FGM. Non-atrophic gastritis was observed in more than half of the background mucosa. All cases were diffusely positive for MUC6 and pepsinogen-I on immunohistochemistry. H+/K+-ATPase staining was negative or showed a scattered pattern in most cases. MUC5AC was expressed on the surface of GA-FGMs. p53 was focally expressed and the Ki67 index was low (1%-20%). Compared with OGA, GA-FG and GA-FGM were more prominent in the macroscopic view (p < 0.05) and had larger sizes (p < 0.0001). Additionally, GA-FG and GA-FGM exhibited higher Ki67 indices than OGA (p < 0.0001). Specimens with Ki-67 proliferation indices >2.5% and size >4.5 mm are more likely to be diagnosed with GA-FG and GA-FGM than OGA. Conclusion: GEN-FGMLs are group of well-differentiated gastric tumors with favourable biological behaviours, low cellular atypia, and low proliferation. Immunohistochemistry is critical for confirming diagnosis. Compared with OGA, GA-FG and GA-FGM have larger sizes and higher Ki67 proliferation indices, indicating that they play a critical role in the identification of GEN-FGML. Pathologists and endoscopists should be cautious to prevent misdiagnosis and overtreatment, especially in biopsy specimens.

Unsupervised Person Re-Identification With Wireless Positioning Under Weak Scene Labeling.

Existing unsupervised person re-identification methods only rely on visual clues to match pedestrians under different cameras. Since visual data is essentially susceptible to occlusion, blur, clothing changes, etc., a promising solution is to introduce heterogeneous data to make up for the defect of visual data. Some works based on full-scene labeling introduce wireless positioning to assist cross-domain person re-identification, but their GPS labeling of entire monitoring scenes is laborious. To this end, we propose to explore unsupervised person re-identification with both visual data and wireless positioning trajectories under weak scene labeling, in which we only need to know the locations of the cameras. Specifically, we propose a novel unsupervised multimodal training framework (UMTF), which models the complementarity of visual data and wireless information. Our UMTF contains a multimodal data association strategy (MMDA) and a multimodal graph neural network (MMGN). MMDA explores potential data associations in unlabeled multimodal data, while MMGN propagates multimodal messages in the video graph based on the adjacency matrix learned from histogram statistics of wireless data. Thanks to the robustness of the wireless data to visual noise and the collaboration of various modules, UMTF is capable of learning a model free of the human label on data. Extensive experimental results conducted on two challenging datasets, i.e., WP-ReID and Campus4K demonstrate the effectiveness of the proposed method.

SignBERT+: Hand-Model-Aware Self-Supervised Pre-Training for Sign Language Understanding.

Hand gesture serves as a crucial role during the expression of sign language. Current deep learning based methods for sign language understanding (SLU) are prone to over-fitting due to insufficient sign data resource and suffer limited interpretability. In this paper, we propose the first self-supervised pre-trainable SignBERT+ framework with model-aware hand prior incorporated. In our framework, the hand pose is regarded as a visual token, which is derived from an off-the-shelf detector. Each visual token is embedded with gesture state and spatial-temporal position encoding. To take full advantage of current sign data resource, we first perform self-supervised learning to model its statistics. To this end, we design multi-level masked modeling strategies (joint, frame and clip) to mimic common failure detection cases. Jointly with these masked modeling strategies, we incorporate model-aware hand prior to better capture hierarchical context over the sequence. After the pre-training, we carefully design simple yet effective prediction heads for downstream tasks. To validate the effectiveness of our framework, we perform extensive experiments on three main SLU tasks, involving isolated and continuous sign language recognition (SLR), and sign language translation (SLT). Experimental results demonstrate the effectiveness of our method, achieving new state-of-the-art performance with a notable gain.

Identification of potential biomarkers associated with CD4+ T cell infiltration in myocardial ischemia-reperfusion injury using bioinformation analysis.

Background: Myocardial ischemia-reperfusion injury (MIRI) is often part of clinical events such as cardiac arrest, resuscitation, and reperfusion after coronary artery occlusion. Recently, more and more studies have shown that the immune microenvironment is an integral part of ischemia-reperfusion injury (IRI), and CD4+ T-cell infiltration plays an important role, but there are no relevant molecular targets for clinical diagnosis and treatment. Methods: The transcriptome data and matched group information were retrieved from the Gene Expression Omnibus (GEO) database. The ImmuCellAI-mouse (Immune Cell Abundance Identifier for mouse) algorithm was used to calculate each symbol's CD4+ T cell infiltration score. The time period with the greatest change in the degree of CD4+ T cell infiltration [ischemia-reperfusion 6 hours (IR6h)-ischemia-reperfusion 24 hours (IR24h)] was selected for the next analysis. Weighted gene co-expression network analysis (WGCNA) and differential expression analysis were performed to screen out CD4+ T cell-related genes and from which the gene CLEC5A was screened for the highest correlation with CD4+ T cell infiltration. The potential regulatory mechanism of CD4+ T cells in MIRI was discussed through various enrichment analysis. Finally, we analyzed the expression and molecular function (MF) of CLEC5A and its related genes in MIRI. Results: A total of 406 CD4+ T cell-related genes were obtained by intersecting the results of WGCNA and differential expression analysis. Functional enrichment analysis indicated that the CD4+ T cell-related genes were mainly involved in chemokine signaling pathway and cell cycle. By constructing a protein-protein interaction (PPI) network, a total of 12 hub genes were identified as candidate genes for further analysis. Through the correlation analysis between the 12 candidate genes found in the PPI network and CD4+ T cell infiltration fraction, we determined the core gene CLEC5A. Finally, a gene interaction network was constructed to decipher the biological functions of CLEC5A using GeneMANIA. Conclusions: In this study, RNA sequencing (RNA-Seq) data at different time points after reperfusion were subjected to a series of bioinformatics methods such as PPI network, WGCNA module, etc., and CLEC5A, a pivotal gene associated with CD4+ T-cells, was found, which may serve as a new target for diagnosis or treatment.

Model-Aware Pre-Training for Radial Distortion Rectification.

Camera lenses often suffer from optical aberrations, causing radial distortion in the captured images. In those images, there exists a clear and general physical distortion model. However, in existing solutions, such rich geometric prior is under-utilized, and the formulation of an effective prediction target is under-explored. To this end, we introduce Radial Distortion TRansformer (RDTR), a new framework for radial distortion rectification. Our RDTR includes a model-aware pre-training stage for distortion feature extraction and a deformation estimation stage for distortion rectification. Technically, on the one hand, we formulate the general radial distortion (i.e., barrel distortion and pincushion distortion) in camera-captured images with a shared geometric distortion model and perform a unified model-aware pre-training for its learning. With the pre-training, the network is capable of encoding the specific distortion pattern of a radially distorted image. After that, we transfer the learned representations to the learning of distortion rectification. On the other hand, we introduce a new prediction target called backward warping flow for rectifying images with any resolution while avoiding image defects. Extensive experiments are conducted on our synthetic dataset, and the results demonstrate that our method achieves state-of-the-art performance while operating in real-time. Besides, we also validate the generalization of RDTR on real-world images. Our source code and the proposed dataset are publicly available at https://github.com/wwd-ustc/RDTR.

Age and treatment disparities in survival of primary malignant cardiac tumors: an analysis of over 40 years and 500 patients.

Background: Primary malignant cardiac tumors (PMCTs) are rare and tend to have a poor prognosis, due to their aggressive biological behavior and the inadequate expertise with the disease. This article compares the survival of patients with PMCT subtypes in the United States across age and treatment groups. Methods: Data of 529 patients diagnosed with PMCTs were analyzed. Chi-squared test was used to assess significance of the differences between proportions in demographic and tumor characteristics by age and treatment. Cox regression analysis was used to estimate survival from the Surveillance, Epidemiology, and End Results (SEER) follow-up data. Results: Survival rates for PMCTs differed significantly between age groups, with patients younger than 20 years surviving significantly longer than those older than 80 years. The median survival times of all patients with PMCTs were 22.5, 11, 5, and 1 month for ages less than 20, 20-50, 51-80, and greater than 80 years, respectively (global log-rank P=0.0026). In the treatment cohort, for all tumors [hazard ratio (HR) 1.52, P<0.001], sarcomas (HR 1.83, P=0.002), and other tumors (HR 2.24, P=0.017), survival was lower in patients who did not receive treatment than in those who received only surgery. Survival after diagnosis of sarcoma was lower in patients who received radiotherapy only than in those who received surgery only (HR 1.49, P=0.046). However, there was no significant association between treatment and survival for lymphoma and mesothelioma. Conclusions: This study confirms that PMCTs have limited treatment options and poor patient survival, especially for elderly patients and patients who receive no treatment. And patients with PMCTs of any age, whether treated or not, have poor survival rates. Techniques for early diagnosis and treatment may be necessary. Surgical treatment should have a higher priority for future treatment of patients with sarcomas.

TransVG++: End-to-End Visual Grounding With Language Conditioned Vision Transformer.

In this work, we explore neat yet effective Transformer-based frameworks for visual grounding. The previous methods generally address the core problem of visual grounding, i.e., multi-modal fusion and reasoning, with manually-designed mechanisms. Such heuristic designs are not only complicated but also make models easily overfit specific data distributions. To avoid this, we first propose TransVG, which establishes multi-modal correspondences by Transformers and localizes referred regions by directly regressing box coordinates. We empirically show that complicated fusion modules can be replaced by a simple stack of Transformer encoder layers with higher performance. However, the core fusion Transformer in TransVG is stand-alone against uni-modal encoders, and thus should be trained from scratch on limited visual grounding data, which makes it hard to be optimized and leads to sub-optimal performance. To this end, we further introduce TransVG++ to make two-fold improvements. For one thing, we upgrade our framework to a purely Transformer-based one by leveraging Vision Transformer (ViT) for vision feature encoding. For another, we devise Language Conditioned Vision Transformer that removes external fusion modules and reuses the uni-modal ViT for vision-language fusion at the intermediate layers. We conduct extensive experiments on five prevalent datasets, and report a series of state-of-the-art records.

Masked Contrastive Representation Learning for Reinforcement Learning.

In pixel-based reinforcement learning (RL), the states are raw video frames, which are mapped into hidden representation before feeding to a policy network. To improve sample efficiency of state representation learning, recently, the most prominent work is based on contrastive unsupervised representation. Witnessing that consecutive video frames in a game are highly correlated, to further improve data efficiency, we propose a new algorithm, i.e., masked contrastive representation learning for RL (M-CURL), which takes the correlation among consecutive inputs into consideration. In our architecture, besides a CNN encoder for hidden presentation of input state and a policy network for action selection, we introduce an auxiliary Transformer encoder module to leverage the correlations among video frames. During training, we randomly mask the features of several frames, and use the CNN encoder and Transformer to reconstruct them based on context frames. The CNN encoder and Transformer are jointly trained via contrastive learning where the reconstructed features should be similar to the ground-truth ones while dissimilar to others. During policy evaluation, the CNN encoder and the policy network are used to take actions, and the Transformer module is discarded. Our method achieves consistent improvements over CURL on 14 out of 16 environments from DMControl suite and 23 out of 26 environments from Atari 2600 Games. The code is available at https://github.com/teslacool/m-curl.

Exploration of the shared genes and signaling pathways between lung adenocarcinoma and idiopathic pulmonary fibrosis.

Background: Idiopathic pulmonary fibrosis (IPF), a type of interstitial lung disease (ILD), is a chronic disease with an unknown etiology. The occurrence of lung cancer (LC) is one of the main causes of death in patients with IPF. However, the pathogenesis driving these malignant transformations remains unclear; therefore, this study aimed to identify the shared genes and functional pathways associated with both disease conditions. Methods: Data were downloaded from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases. To identify overlapping genes in both diseases, the "limma" package in R software and weighted gene coexpression network analysis (WGCNA) were used. Venn diagrams were used to obtain the shared genes. The diagnostic value of the shared genes was assessed using receiver operating characteristic (ROC) curve analysis. Gene Ontology (GO) term enrichment was performed on the shared genes between lung adenocarcinoma (LUAD) and IPF, and the genes were also functionally enriched using Metascape. A protein-protein interaction (PPI) network was created using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database. Finally, the link between shared genes and common antineoplastic medicines was investigated using the CellMiner database. Results: The coexpression modules associated with LUAD and IPF were discovered using WGCNA, and 148 genes were found to overlap. In addition, 74 upregulated and 130 downregulated overlapping genes were obtained via differential gene analysis. Functional analysis of the genes revealed that these genes are primarily engaged in extracellular matrix (ECM) pathways. Furthermore, COL1A2, POSTN, COL5A1, CXCL13, CYP24A1, CXCL14, and BMP2 were identified as potential biomarkers in patients with LUAD secondary to IPF showing good diagnostic values. Conclusions: ECM-related mechanisms may be the underlying link between LC and IPF. A total of 7 shared genes were identified as potential diagnostic markers and therapeutic targets for LUAD and IPF.

Identification of a polycomb group-related gene signature for predicting prognosis and immunotherapy efficacy in lung adenocarcinoma.

Background: Several studies have reported the role of polycomb group (PcG) genes in human cancers; however, their role in lung adenocarcinoma (LUAD) is unknown. Methods: Firstly, consensus clustering analysis was used to identify PcG patterns among the 633 LUAD samples in the training dataset. The PcG patterns were then compared in terms of the overall survival (OS), signaling pathway activation, and immune cell infiltration. The PcG-related gene score (PcGScore) was developed using Univariate Cox regression and the least absolute shrinkage and selection operator (LASSO) algorithm to estimate the prognostic value and treatment sensitivity of LUAD. Finally, the prognostic ability of the model was validated using a validation dataset. Results: Two PcG patterns were obtained by consensus clustering analysis, and the two patterns showed significant differences in prognosis, immune cell infiltration, and signaling pathways. Both the univariate and multivariate Cox regression analyses confirmed that the PcGScore was a reliable and independent predictor of LUAD (P<0.001). The high- and low-PCGScore groups showed significant differences in the prognosis, clinical outcomes, genetic variation, immune cell infiltration, and immunotherapeutic and chemotherapeutic effects. Lastly, the PcGScore demonstrated exceptional accuracy in predicting the OS of the LUAD patients in a validation dataset (P<0.001). Conclusions: The study indicated that the PcGScore could serve as a novel biomarker to predict prognosis, clinical outcomes, and treatment sensitivity for LUAD patients.

Identification of a potential bioinformatics-based biomarker in keloids and its correlation with immune infiltration.

Keloid formation is a pathological consequence resulting from cutaneous irritation and injury, primarily attributed to excessive collagen matrix deposition and fibrous tissue proliferation. Chronic inflammation, left uncontrolled over an extended period, also stands as a substantial contributing factor. The precise mechanisms underlying keloid formation remain unclear. Therefore, this study aimed to identify key genes for diagnostic purposes. To achieve this, we used two Gene Expression Omnibus (GEO) data sets to identify differentially expressed genes. We identified one particular gene, homeobox C9 (HOXC9), using a thorough strategy involving two algorithms (least absolute shrinkage and selection operator and support vector machine-recursive feature elimination) and weighted gene co-expression network analysis. We then assessed its expression in normal and keloid tissues. In addition, we explored its temporal expression patterns via Mfuzz time clustering analysis. In our comprehensive analysis, we observed that immune infiltration, as well as cell proliferation, are crucial to keloid formation. Thus, we investigated immune cell infiltration in the keloid and normal groups, as well as the correlation between HOXC9 and these immune cells. It was found that HOXC9 was closely associated with the immune microenvironment of keloids. This shows that HOXC9 can serve as a potential biomarker and therapeutic target for keloids.

Identification of lysosomal genes associated with prognosis in lung adenocarcinoma.

Background: Lung adenocarcinoma (LUAD) is the most common subtype of lung cancer, representing 40% of all cases of this tumor. Despite immense improvements in understanding the molecular basis, diagnosis, and treatment of LUAD, its recurrence rate is still high. Methods: RNA-seq data from The Cancer Genome Atlas (TCGA) LUAD cohort were download from Genomic Data Commons Portal. The GSE13213 dataset from Gene Expression Omnibus (GEO) was used for external validation. Differential prognostic lysosome-related genes (LRGs) were identified by overlapping survival-related genes obtained via univariate Cox regression analysis with differentially expressed genes (DEGs). The prognostic model was built using Kaplan-Meier curves and least absolute shrinkage and selection operator (LASSO) analyses. In addition, univariate and multivariate Cox analyses were employed to identify independent prognostic factors. The responses of patients to immune checkpoint inhibitors (ICIs) were further predicted. The pRRophetic package and rank-sum test were used to compute the half maximal inhibitory concentrations (IC50) of 56 chemotherapeutic drugs and their differential effects in the low- and high-risk groups. Moreover, quantitative real-time polymerase chain reaction, Western blot, and human protein atlas (HPA) database were used to verify the expression of the four prognostic biomarkers in LUAD. Results: Of the nine candidate differential prognostic LRGs, GATA2, TFAP2A, LMBRD1, and KRT8 were selected as prognostic biomarkers. The prediction of the risk model was validated to be reliable. Cox independent prognostic analysis revealed that risk score and stage were independent prognostic factors in LUAD. Furthermore, the nomogram and calibration curves of the independent prognostic factors performed well. Differential analysis of ICIs revealed CD276, ICOS, PDCD1LG2, CD27, TNFRSF18, TNFSF9, ENTPD1, and NT5E to be expressed differently in the low- and high-risk groups. The IC50 values of 12 chemotherapeutic drugs, including epothilone.B, JNK.inhibitor.VIII, and AKT.inhibitor.VIII, significantly differed between the two risk groups. KRT8 and TFAP2A were highly expressed, while GATA2 and LMBRD1 were poorly expressed in LUAD cell lines. In addition, KRT8 and TFAP2A were highly expressed, while GATA2 and LMBRD1 were poorly expressed in tumor tissues. Conclusions: Four key prognostic biomarkers-GATA2, TFAP2A, LMBRD1, and KRT8-were used to construct a significant prognostic model for LUAD patients.

End-to-End Optimized Versatile Image Compression With Wavelet-Like Transform.

Built on deep networks, end-to-end optimized image compression has made impressive progress in the past few years. Previous studies usually adopt a compressive auto-encoder, where the encoder part first converts image into latent features, and then quantizes the features before encoding them into bits. Both the conversion and the quantization incur information loss, resulting in a difficulty to optimally achieve arbitrary compression ratio. We propose iWave++ as a new end-to-end optimized image compression scheme, in which iWave, a trained wavelet-like transform, converts images into coefficients without any information loss. Then the coefficients are optionally quantized and encoded into bits. Different from the previous schemes, iWave++ is versatile: a single model supports both lossless and lossy compression, and also achieves arbitrary compression ratio by simply adjusting the quantization scale. iWave++ also features a carefully designed entropy coding engine to encode the coefficients progressively, and a de-quantization module for lossy compression. Experimental results show that lossy iWave++ achieves state-of-the-art compression efficiency compared with deep network-based methods; on the Kodak dataset, lossy iWave++ leads to 17.34 percent bits saving over BPG; lossless iWave++ achieves comparable or better performance than FLIF. Our code and models are available at https://github.com/mahaichuan/Versatile-Image-Compression.

Survival of gastric cancer in China from 2000 to 2022: A nationwide systematic review of hospital-based studies.

Background: Gastric cancer (GC) mortality continues to fall in industrialized countries, but still remains a public health concern in China, accounting for more than 370 000 deaths. We aimed to evaluate the survival of GC in China from 2000 to 2022 through a nationwide systematic review of hospital-based studies and to identify whether hospital-based studies show higher survival rates than population-based studies. Methods: We searched PubMed, Embase, Web of Science, and the Chinese databases of CNKI and Wanfang for hospital-based studies on GC survival published between January 1, 2000, and January 20, 2022. We calculated the nationwide GC survival rate (SR) and its 95% confidence interval (CI) and conducted subgroup analyses on histologic type, subsite, tumour node metastasis (TNM) stage, therapy type, study design, and participant region. The study protocol was registered in PROSPERO (CRD-42019121559). Results: The initial literature search returned 36 613 publications, among which 664 studies (180 798 participants) matched the inclusion criteria and were included in the meta-analysis. The pooled one-, two-, three- and five-year SRs of GC were 75.4% (95% CI = 74.0%-76.8%), 54.3% (95% CI = 50.1%-58.6%), 53.4% (95% CI = 50.4%-56.4%), and 44.5% (95% CI = 41.5%-47.5%), respectively. Subgroup analyses revealed an increase in three- and five-year SRs from 2006 to 2022. The five-year SR was highest among patients without lymph node metastasis (pooled SR = 67.8%, 95% CI = 62.8%-72.7%) and lowest among those with distant metastasis (pooled SR = 8.4%, 95% CI = 5.1%-11.7%). Conclusions: Our findings illustrate that the long-term survival of GC has improved in China since 2000. Hospital-based studies have presented higher SRs than population-based surveillance.

Passive Non-Line-of-Sight Imaging Using Optimal Transport.

Passive non-line-of-sight (NLOS) imaging has drawn great attention in recent years. However, all existing methods are in common limited to simple hidden scenes, low-quality reconstruction, and small-scale datasets. In this paper, we propose NLOS-OT, a novel passive NLOS imaging framework based on manifold embedding and optimal transport, to reconstruct high-quality complicated hidden scenes. NLOS-OT converts the high-dimensional reconstruction task to a low-dimensional manifold mapping through optimal transport, alleviating the ill-posedness in passive NLOS imaging. Besides, we create the first large-scale passive NLOS imaging dataset, NLOS-Passive, which includes 50 groups and more than 3,200,000 images. NLOS-Passive collects target images with different distributions and their corresponding observed projections under various conditions, which can be used to evaluate the performance of passive NLOS imaging algorithms. It is shown that the proposed NLOS-OT framework achieves much better performance than the state-of-the-art methods on NLOS-Passive. We believe that the NLOS-OT framework together with the NLOS-Passive dataset is a big step and can inspire many ideas towards the development of learning-based passive NLOS imaging. Codes and dataset are publicly available (https://github.com/ruixv/NLOS-OT).

Histone acetylation modification regulator-mediated tumor microenvironment infiltration characteristics and prognostic model of lung adenocarcinoma patients.

Background: The incidence rate of lung adenocarcinoma (LUAD) is rapidly increasing. Recent studies have reported that histone acetylation modification plays an important role in the occurrence and development of tumors. However, the potential role of modification of histone acetylation modification in the development of tumor immune microenvironment is still unclear. Methods: In this study, we comprehensively evaluated the acetylation modification patterns of LUAD samples obtained from various different databases based on 36 histone modification regulators, and constructed a prognostic model based on The Cancer Genome Atlas (TCGA) LUAD cohort using the Cox regression method. The close relationship between histone acetylation and tumor immune characteristics was further studied, including immune infiltration, immune escape and immunotherapy. Finally, we combined three cohort (GSE30219, GSE72094 and GSE50081) from Gene Expression Omnibus (GEO) database to verify the above results. Results: We analyzed the expression, mutation and interaction of 36 histone acetylation regulated genes. After Univariate Cox regression analysis and least absolute shrinkage and selection operator regression (LASSO), 5 genes (KAT2B, SIRT2, HDAC5, KAT8, HDAC2) were screened to establish the prognosis model and calculate the risk score. Then, patients in the TCGA cohort were divided into high- and low-risk groups based on the risk scores. Further analysis indicated that patients in the high-risk group exhibited significantly reduced overall survival (OS) compared with those in the low-risk group. The high- and low-risk groups exhibited significant differences in terms of tumor immune characteristics, such as immune infiltration, immune escape and immunotherapy. The high-risk group had lower immune score, less immune cell infiltration and higher clinical stage. Moreover, multivariate analysis revealed that this prognostic model might be a powerful prognostic predictor for LUAD. In addition, drugs sensitive for this classification were identified. Finally, the efficacy of the prognostic model was validated by cohort (GSE30219, GSE72094 and GSE50081) from GEO database. Conclusions: Our study provided a robust signature for predicting changing prognosis of patients with LUAD. Thus, it appears to be a potentially useful prognostic tool. Moreover, the important relationship between histone acetylation and tumor immune microenvironment was revealed.