The effects of meteorological factors and air pollutants on the incidence of tuberculosis in people living with HIV/AIDS in subtropical Guangxi, China.

BACKGROUND: Previous studies have shown the association between tuberculosis (TB) and meteorological factors/air pollutants. However, little information is available for people living with HIV/AIDS (PLWHA), who are highly susceptible to TB. METHOD: Data regarding TB cases in PLWHA from 2014 to2020 were collected from the HIV antiviral therapy cohort in Guangxi, China. Meteorological and air pollutants data for the same period were obtained from the China Meteorological Science Data Sharing Service Network and Department of Ecology and Environment of Guangxi. A distribution lag non-linear model (DLNM) was used to evaluate the effects of meteorological factors and air pollutant exposure on the risk of TB in PLWHA. RESULTS: A total of 2087 new or re-active TB cases were collected, which had a significant seasonal and periodic distribution. Compared with the median values, the maximum cumulative relative risk (RR) for TB in PLWHA was 0.663 (95% confidence interval [CI]: 0.507-0.866, lag 4 weeks) for a 5-unit increase in temperature, and 1.478 (95% CI: 1.116-1.957, lag 4 weeks) for a 2-unit increase in precipitation. However, neither wind speed nor PM10 had a significant cumulative lag effect. Extreme analysis demonstrated that the hot effect (RR = 0.638, 95%CI: 0.425-0.958, lag 4 weeks), the rainy effect (RR = 0.285, 95%CI: 0.135-0.599, lag 4 weeks), and the rainless effect (RR = 0.552, 95%CI: 0.322-0.947, lag 4 weeks) reduced the risk of TB. Furthermore, in the CD4(+) T cells < 200 cells/µL subgroup, temperature, precipitation, and PM10 had a significant hysteretic effect on TB incidence, while temperature and precipitation had a significant cumulative lag effect. However, these effects were not observed in the CD4(+) T cells ≥ 200 cells/µL subgroup. CONCLUSION: For PLWHA in subtropical Guangxi, temperature and precipitation had a significant cumulative effect on TB incidence among PLWHA, while air pollutants had little effect. Moreover, the influence of meteorological factors on the incidence of TB also depends on the immune status of PLWHA.

Effectiveness of human immunodeficiency virus prevention strategies by mapping the geographic dispersion pattern of human immunodeficiency virus prevalence in Nanning, China.

BACKGROUND: The Guangxi government initiated two rounds of the Guangxi AIDS Conquering Project (GACP) in 2010 (Phase I) and 2015 (Phase II) to control human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) epidemics. However, the effectiveness of GACP in HIV prevention and treatment has rarely been reported. This study aimed to assess the effectiveness of the GACP implemented in Guangxi, China and provide data for strategy and praxis improvements to achieve Joint United Nations Programme on HIV/AIDS (UNAIDS) 95-95 targets. METHODS: We used spatial approaches to trace the spatiotemporal distribution properties, epidemic trends, and correlation between macroscopic factors and HIV incidence using data from the Chinese HIV/AIDS case reporting system to explore the effects of the GACP. RESULTS: During the GACP era, the HIV epidemic stabilized in urban centers, showing a downward trend in the Hengzhou and Binyang Counties in the eastern region, whereas it continued to increase in rural areas of the northwest region, such as the Long'an, Mashan, Shanglin, and Wuming Districts. The linear directional mean (LDM) of HIV infection reported cases displayed a southeast-northwest direction, with an LDM value of 12.52°. Compared with that in Phase I, Hengzhou withdrew from the high-high clustering area, and the west-north suburban counties pulled out the low-low clustering area during Phase II. Significant HIV clusters were identified in the eastern region during Phase I, whereas these clusters emerged in the northwestern areas during Phase II. Regarding HIV, socioeconomic status, population mobility, and medical care levels were the key social drivers of heterogeneous spatial distribution. CONCLUSIONS: The GACP assisted in effectively managing the HIV epidemic in urban and eastern areas of Nanning City. However, prevention and control efforts in rural regions, particularly those located in the northwest, may not have yielded comparable outcomes. To address this disparity, allocating additional resources and implementing tailored intervention measures for these rural areas are imperative.

Near Full-Length Genome Characterization of Two Novel Unique Recombinants (CRF01_AE/CRF07_BC) in Beijing, China.

With the prevalence of human immunodeficiency virus type 1 (HIV-1) CRF01_AE and CRF07_BC subtypes in China, the co-circulation of multiple subtypes in the HIV-1-positive population may result in dual infection or superinfection in the population, leading to the emergence of unique recombinant forms (URFs) of the HIV-1 virus. In this study, two second-generation unique recombinant strains, BI0114 and BI0116, were identified, and their near full-length genome sequences were obtained. Recombination analysis showed that both sequences were isoforms of URF_0107, and they were second-generation unique recombinant strains formed by the recombination of CRF01_AE and CRF07_BC, with the isoforms being CRF01_AE and CRF0107_BC, respectively. The continued emergence of novel CRF01_AE/CRF07_BC recombinant strains suggests that the epidemiological, preventive, and control situation of HIV-1 is complex and that the relevant health authorities urgently need to establish responses to the challenges posed by changes in the pattern of strain recombination.

The origin, dissemination, and molecular networks of HIV-1 CRF65_cpx strain in Hainan Island, China.

BACKGROUND: HIV-1 CRF65_cpx strain carries drug-resistant mutations, which raises concerns about its potential for causing virologic failure. The CRF65_cpx ranks as the fourth most prevalent on Hainan Island, China. However, the origin and molecular epidemiology of CRF65_cpx strains in this area remain unclear. This study aims to estimate the spatial origins and dissemination patterns of HIV-1 CRF65_cpx in this specific region. METHODS: Between 2018 and 2021, a total of 58 pol sequences of the CRF65_cpx were collected from HIV-positive patients on Hainan Island. The available CRF65_cpx pol sequences from public databases were compiled. The HIV-TRACE tool was used to construct transmission networks. The evolutionary history of the introduction and dissemination of HIV-1 CRF65_cpx on Hainan Island were analyzed using phylogenetic analysis and the Bayesian coalescent-based approach. RESULTS: Among the 58 participants, 89.66% were men who have sex with men (MSM). The median age was 25 years, and 43.10% of the individuals had a college degree or above. The results indicated that 39 (67.24%) sequences were interconnected within a single transmission network. A consistent expansion was evident from 2019 to 2021, with an incremental annual addition of four sequences into the networks. Phylodynamic analyses showed that the CRF65_cpx on Hainan Island originated from Beijing (Bayes factor, BF = 17.4), with transmission among MSM on Hainan Island in 2013.2 (95%HPD: 2012.4, 2019.5), subsequently leading to an outbreak. Haikou was the local center of the CRF65_cpx epidemic. This strain propagated from Haikou to other locations, including Sanya (BF > 1000), Danzhou (BF = 299.3), Chengmai (BF = 27.0) and Tunchang (BF = 16.3). The analyses of the viral migration patterns between age subgroups and risk subgroups revealed that the viral migration directions were from "25-40 years old" to "17-24 years old" (BF = 14.6) and to "over 40 years old" (BF = 17.6), and from MSM to heterosexuals (BF > 1000) on Hainan Island. CONCLUSION: Our analyses elucidate the transmission dynamics of CRF65_cpx strain on Hainan Island. Haikou is identified as the potential hotspot for CRF65_cpx transmission, with middle-aged MSM identified as the key population. These findings suggest that targeted interventions in hotspots and key populations may be more effective in controlling the HIV epidemic.

Willingness to accept monkeypox vaccine and its correlates among men who have sex with men in Southern China: a web-based online cross-sectional study.

Background: The May 2022 global outbreak of monkeypox (MPX) poses a threat to the health of men who have sex with men. However, there is limited data on the willingness of MSM to receive monkeypox vaccination in Southern China. This study aimed to assess the knowledge of MPX, concerns regarding MPX, and willingness to receive monkeypox vaccination, as well as their correlates, among MSM in China. Methods: We conducted a Web-based online survey of MSM in Southern China from August to September 2022. Data were collected on the socio-demographic characteristics, knowledge, worries, concerns regarding MPX and willingness to receive monkeypox vaccination. Multivariate logistic regression was employed to explore the factors associated with willingness to receive monkeypox vaccination. Results: A total of 1903 participants completed the survey. Among them, approximately 69.9% reported being aware of MPX awareness, 94.1% of the participants supported the promotion of monkeypox vaccination. The majority of participants (91.4%) expressed their willingness to receive monkeypox vaccination. Participants who considered monkeypox vaccination safe [adjusted odds ratio (aOR) = 4.82, 95% CI: 1.35-17.18], agreed on the necessity of government promotion of monkeypox vaccination in China (aOR = 6.03, 95% CI: 1.07-33.93), believed in prioritizing monkeypox vaccination for MSM (aOR = 5.01, 95% CI: 1.10-22.71), and had friends or sexual partners who had already received the monkeypox or smallpox vaccination (aOR = 10.37, 95% CI: 2.11-50.99) are more likely to be vaccinated. Conversely, married individuals (aOR = 0.13, 95% CI: 0.03-0.47), those engaging in anal sex 4-6 times per week in the past 3 months (aOR = 0.26, 95% CI: 0.09-0.77) expressed hesitancy toward monkeypox vaccination. Conclusion: There was a high willingness to receive monkeypox vaccination among MSM in China. The hesitancy toward the monkeypox vaccine can be effectively mitigated by addressing concerns about its safety and potential adverse reactions. Moreover, increasing acceptance of the monkeypox vaccination among MSM and their peers is crucial, as social influence significantly impacts vaccine attitudes and behaviors.

Dynamic m6 A profiles reveal the role of YTHDC2-TLR2 signaling axis in Talaromyces marneffei infection.

Talaromyces marneffei (TM) immune evasion is an important factor leading to the high mortality rate of Penicilliosis marneffei. N6 -methyladenosine (m6 A) plays important roles in host immune response to various pathogen infections, yet its role in TM and HIV/TM coinfection remains largely unexplored. Here we reported genome-wide transcriptional m6 A profiles of TM mono-infection and HIV/TM coinfection. Our finding revealed dynamic alterations in global m6 A levels and upregulation of the m6 A reader YTH N6 -methyladenosine RNA binding protein C2 (YTHDC2) in TM-infected macrophages. Knockdown of YTHDC2 in TM-infected cells showed an elevated expression of TLR2 through m6 A-dependence, along with upregulation of TNF-α and IL1-ß. Overall, we characterized the m6 A profiles of the host and fungus before and after TM infection, and demonstrated that YTHDC2 mediates the key m6 A site of TLR2 to exert its function. These findings provide new insights into the underlying mechanisms and novel therapeutic approaches for TM diseases.

Image Deblurring by Exploring In-Depth Properties of Transformer.

Image deblurring continues to achieve impressive performance with the development of generative models. Nonetheless, there still remains a displeasing problem if one wants to improve perceptual quality and quantitative scores of recovered image at the same time. In this study, drawing inspiration from the research of transformer properties, we introduce the pretrained transformers to address this problem. In particular, we leverage deep features extracted from a pretrained vision transformer (ViT) to encourage recovered images to be sharp without sacrificing the performance measured by the quantitative metrics. The pretrained transformer can capture the global topological relations (i.e., self-similarity) of image, and we observe that the captured topological relationships about the sharp image will change when blur occurs. By comparing the transformer features between recovered image and target one, the pretrained transformer provides high-resolution blur-sensitive semantic information, which is critical in measuring the sharpness of the deblurred image. On the basis of the advantages, we present two types of novel perceptual losses to guide image deblurring. One regards the features as vectors and computes the discrepancy between representations extracted from recovered image and target one in Euclidean space. The other type considers the features extracted from an image as a distribution and compares the distribution discrepancy between recovered image and target one. We demonstrate the effectiveness of transformer properties in improving the perceptual quality while not sacrificing the quantitative scores peak signal-to-noise ratio (PSNR) over the most competitive models, such as Uformer, Restormer, and NAFNet, on defocus deblurring and motion deblurring tasks. The code is available at https://github. com/erfect2020/TransformerPerceptualLoss.

Causal relationship between gut microbiota and tuberculosis: a bidirectional two-sample Mendelian randomization analysis.

BACKGROUND: Growing evidence from observational studies and clinical trials suggests that the gut microbiota is associated with tuberculosis (TB). However, it is unclear whether any causal relationship exists between them and whether causality is bidirectional. METHODS: A bidirectional two-sample Mendelian randomization (MR) analysis was performed. The genome-wide association study (GWAS) summary statistics of gut microbiota were obtained from the MiBioGen consortium, while the GWAS summary statistics of TB and its specific phenotypes [respiratory tuberculosis (RTB) and extrapulmonary tuberculosis (EPTB)] were retrieved from the UK Biobank and the FinnGen consortium. And 195 bacterial taxa from phylum to genus were analyzed. Inverse variance weighted (IVW), MR-Egger regression, maximum likelihood (ML), weighted median, and weighted mode methods were applied to the MR analysis. The robustness of causal estimation was tested using the heterogeneity test, horizontal pleiotropy test, and leave-one-out method. RESULTS: In the UK Biobank database, we found that 11 bacterial taxa had potential causal effects on TB. Three bacterial taxa genus.Akkermansia, family.Verrucomicrobiacea, order.Verrucomicrobiales were validated in the FinnGen database. Based on the results in the FinnGen database, the present study found significant differences in the characteristics of gut microbial distribution between RTB and EPTB. Four bacterial taxa genus.LachnospiraceaeUCG010, genus.Parabacteroides, genus.RuminococcaceaeUCG011, and order.Bacillales were common traits in relation to both RTB and TB, among which order.Bacillales showed a protective effect. Additionally, family.Bacteroidacea and genus.Bacteroides were identified as common traits in relation to both EPTB and TB, positively associating with a higher risk of EPTB. In reverse MR analysis, no causal association was identified. No significant heterogeneity of instrumental variables (IVs) or horizontal pleiotropy was found. CONCLUSION: Our study supports a one-way causal relationship between gut microbiota and TB, with gut microbiota having a causal effect on TB. The identification of characteristic gut microbiota provides scientific insights for the potential application of the gut microbiota as a preventive, diagnostic, and therapeutic tool for TB.

A bibliometric analysis of m6A methylation in viral infection from 2000 to 2022.

BACKGROUND: N6-methyladenosine (m6A) methylation has become an active research area in viral infection, while little bibliometric analysis has been performed. In this study, we aim to visualize hotspots and trends using bibliometric analysis to provide a comprehensive and objective overview of the current research dynamics in this field. METHODS: The data related to m6A methylation in viral infection were obtained through the Web of Science Core Collection form 2000 to 2022. To reduce bias, the literature search was conducted on December 1, 2022. Bibliometric and visual analyzes were performed using CiteSpace and Bibliometrix package. After screening, 319 qualified records were retrieved. RESULTS: These publications mainly came from 28 countries led by China and the United States (the US), with the US ranking highest in terms of total link strength.The most common keywords were m6A, COVID-19, epitranscriptomics, METTL3, hepatitis B virus, innate immunity and human immunodeficiency virus 1. The thematic map showed that METTL3, plant viruses, cancer progression and type I interferon (IFN-I) reflected a good development trend and might become a research hotspot in the future, while post-transcriptional modification, as an emerging or declining theme, might not develop well. CONCLUSIONS: In conclusion, m6A methylation in viral infection is an increasingly important topic in articles. METTL3, plant viruses, cancer progression and IFN-I may still be research hotspots and trends in the future.

The role and mechanisms of PD-L1 in immune evasion during Talaromyces marneffei infection.

Talaromycosis, caused by Talaromyces marneffei (T. marneffei), is a systemic fungal disease that involves dissemination throughout the body. The ability of T. marneffei to evade the immune system is considered a crucial factor in its persistent infection, although the specific mechanisms are not yet fully understood. This study aims to investigate the molecular mechanisms underlying the occurrence of latent T. marneffei infection and immune evasion. The gene expression profile analysis in T. marneffei-infected mouse revealed that Pd-l1 exhibited the highest correlation strength with other hub genes, with a median of 0.60 (IQR: 0.50-0.69). T. marneffei infection upregulated the expression of PD-1 and PD-L1 in PBMCs from HIV patients, which was also observed in the T. marneffei-infected mouse and macrophage models. Treatment with a PD-L1 inhibitor significantly reduced fungal burden in the liver and spleen tissues of infected mice and in the kupffer-CTLL-2 co-culture system. PD-L1 inhibitor treatment increased CTLL-2 cell proliferation and downregulated the expression of PD-1, SHP-2, and p-SHP-2, indicating the activation of T cell viability and T cell receptor signaling pathway. Additionally, treatment with a PI3K inhibitor downregulated PD-L1 in T. marneffei-infected kupffer cells. Similar results were observed with treatment using the T. marneffei cell wall virulence factor ß-glucan. Overall, T. marneffei infection upregulated PD-L1 expression in HIV / T. marneffei patients, mice, and kupffer cells. Treatment with a PD-L1 inhibitor significantly reduced fungal burden, while activating T cell activity and proliferation, thereby promoting fungal clearance. Furthermore, the PI3K signaling pathway may be involved in the regulation of PD-L1 by T. marneffei.

Talaromyces marneffei suppresses macrophage inflammation by regulating host alternative splicing.

Talaromyces marneffei (T. marneffei) immune escape is essential in the pathogenesis of talaromycosis. It is currently known that T. marneffei achieves immune escape through various strategies. However, the role of cellular alternative splicing (AS) in immune escape remains unclear. Here, we depict the AS landscape in macrophages upon T. marneffei infection via high-throughput RNA sequencing and detect a truncated protein of NCOR2 / SMRT, named NCOR2-013, which is significantly upregulated after T. marneffei infection. Mechanistic analysis indicates that NCOR2-013 forms a co-repression complex with TBL1XR1 / TBLR1 and HDAC3, thereby inhibiting JunB-mediated transcriptional activation of pro-inflammatory cytokines via the inhibition of histone acetylation. Furthermore, we identify TUT1 as the AS regulator that regulates NCOR2-013 production and promotes T. marneffei immune evasion. Collectively, these findings indicate that T. marneffei escapes macrophage killing through TUT1-mediated alternative splicing of NCOR2 / SMRT, providing insight into the molecular mechanisms of T. marneffei immune evasion and potential targets for talaromycosis therapy.

Characteristics and risk factors for readmission in HIV-infected patients with Talaromyces marneffei infection.

OBJECTIVES: Talaromyces marneffei (T. marneffei) is an opportunistic fungal infection (talaromycosis), which is common in subtropical regions and is a leading cause of death in HIV-1-infected patients. This study aimed to determine the characteristics and risk factors associated with hospital readmissions in HIV patients with T. marneffei infection in order to reduce readmissions. METHODS: We conducted a retrospective study of admitted HIV-infected individuals at the Fourth People's Hospital of Nanning, Guangxi, China, from 2012 to 2019. Kaplan-Meier analyses and Principal component analysis (PCA) were used to evaluate the effects of T. marneffei infection on patient readmissions. Additionally, univariate and multifactorial analyses, as well as Propensity score matching (PSM) were used to analyze the factors associated with patient readmissions. RESULTS: HIV/AIDS patients with T. marneffei-infected had shorter intervals between admissions and longer lengths of stay than non-T. marneffei-infected patients, despite lower readmission rates. Compared with non-T. marneffei-infected patients, the mortality rate for talaromycosis patients was higher at the first admission. Among HIV/AIDS patients with opportunistic infections, the mortality rate was highest for T. marneffei at 16.2%, followed by cryptococcus at 12.5%. However, the readmission rate was highest for cryptococcus infection (37.5%) and lowest for T. marneffei (10.8%). PSM and Logistic regression analysis identified leukopenia and elevated low-density lipoprotein (LDL) as key factors in T.marneffei-infected patients hospital readmissions. CONCLUSIONS: The first admission represents a critical window to intervene in the prognosis of patients with T. marneffei infection. Leukopenia and elevated LDL may be potential risk factors impacting readmissions. Our findings provide scientific evidence to improve the long-term outcomes of HIV patients with T. marneffei infection.

Biological Characteristics and Functional Analysis of the Linoleic Acid Synthase Gene ZjFAD2 in Jujube.

Jujube fruit is rich in linoleic acid and other bioactive components and has great potential to be used for the development of functional foods. However, the roles of FAD2 genes in linoleic acid biosynthesis in jujube fruit remain unclear. Here, we identified 15 major components in jujube and found that linoleic acid was the main unsaturated fatty acid; major differences in the content and distribution of linoleic acid in the pulp and seeds were observed, and levels of linoleic acid decreased during fruit maturation. Analysis of the fatty acid metabolome, genome, and gene expression patterns of cultivated and wild-type jujube revealed five ZjFAD2 family members highly related to linoleic acid biosynthesis. The heterologous expression of these five ZjFAD2 family members in tobacco revealed that all five of these genes increased the content of linoleic acid. Additionally, transient expression of these genes in jujube fruit and the virus-induced gene silencing (VIGS) test further confirmed the key roles of ZjFAD2-11 and ZjFAD2-1 in the biosynthesis of linoleic acid. The results of this research provide valuable insights into the molecular mechanism underlying linoleic acid synthesis in jujube and will aid the development of quality-oriented breeding strategies.

Activation of the JNK/COX-2/HIF-1α axis promotes M1 macrophage via glycolytic shift in HIV-1 infection.

Chronic inflammation is recognized as a major risk factor for the severity of HIV infection. Whether metabolism reprogramming of macrophages caused by HIV-1 is related to chronic inflammatory activation, especially M1 polarization of macrophages, is inconclusive. Here, we show that HIV-1 infection induces M1 polarization and enhanced glycolysis in macrophages. Blockade of glycolysis inhibits M1 polarization of macrophages, indicating that HIV-1-induced M1 polarization is supported by enhanced glycolysis. Moreover, we find that this immunometabolic adaptation is dependent on hypoxia-inducible factor 1α (HIF-1α), a strong inducer of glycolysis. HIF-1α-target genes, including HK2, PDK1, and LDHA, are also involved in this process. Further research discovers that COX-2 regulates HIF-1α-dependent glycolysis. However, the elevated expression of COX-2, enhanced glycolysis, and M1 polarization of macrophages could be reversed by inactivation of JNK in the context of HIV-1 infection. Our study mechanistically elucidates that the JNK/COX-2/HIF-1α axis is activated to strengthen glycolysis, thereby promoting M1 polarization in macrophages in HIV-1 infection, providing a new idea for resolving chronic inflammation in clinical AIDS patients.

Construction of a murine model of latent infection and reactivation induced by Talaromyces marneffei.

OBJECTIVE: To establish a murine model of Talaromyces marneffei (T. marneffei) latent infection and reactivation, providing a foundation for exploring the molecular mechanisms underlying disease relapse. METHODS: BALB/c mice were tail vein injected with T. marneffei at 0 days post-infection (dpi) and treated with cyclophosphamide (CTX) intraperitoneally every four days, starting from 21 dpi or 42 dpi. Mice were observed for body weight changes, liver and spleen indices, histological characteristics of liver and spleen, fungal load detection in liver and spleen, and Mp1p qualitation in liver and spleen to assess T. marneffei infection severity. RESULTS: T. marneffei-infected mice exhibited a trend of initial weight loss followed by recovery and a subsequent decrease in weight after CTX injection throughout the observation period. Liver and spleen indices, as well as tissue damage, significantly increased during infection but later returned to normal levels, with a gradual rise observed after immunosuppression. Fungal load analysis revealed positive T. marneffei cultures in the liver and spleen at 7 dpi and 14 dpi, followed by negative T. marneffei cultures from 21 dpi until day 21 post-immunosuppression (42 dpi or 63 dpi); however, the spleen remained T. marneffei-cultured negative, consistent with the trend observed in Mp1p detection results. CONCLUSION: A latent infection and reactivation model of T. marneffei in mice was successfully established, with the liver likely serving as a key site for latent T. marneffei.

Unsupervised Deep Exemplar Colorization via Pyramid Dual Non-Local Attention.

Exemplar-based colorization is a challenging task, which attempts to add colors to the target grayscale image with the aid of a reference color image, so as to keep the target semantic content while with the reference color style. In order to achieve visually plausible chromatic results, it is important to sufficiently exploit the global color style and the semantic color information of the reference color image. However, existing methods are either clumsy in exploiting the semantic color information, or lack of the dedicated fusion mechanism to decorate the target grayscale image with the reference semantic color information. Besides, these methods usually use a single-stage encoder-decoder architecture, which results in the loss of spatial details. To remedy these problems, we propose an effective exemplar colorization strategy based on pyramid dual non-local attention network to exploit the long-range dependency as well as multi-scale correlation. Specifically, two symmetrical branches of pyramid non-local attention block are tailored to achieve alignments from the target feature to the reference feature and from the reference feature to the target feature respectively. The bidirectional non-local fusion strategy is further applied to get a sufficient fusion feature that achieves full semantic consistency between multi-modal information. To train the network, we propose an unsupervised learning manner, which employs the hybrid supervision including the pseudo paired supervision from the reference color images and unpaired supervision from both the target grayscale and reference color images. Extensive experimental results are provided to demonstrate that our method achieves better photo-realistic colorization performance than the state-of-the-art methods.

A Practical Contrastive Learning Framework for Single-Image Super-Resolution.

Contrastive learning has achieved remarkable success on various high-level tasks, but there are fewer contrastive learning-based methods proposed for low-level tasks. It is challenging to adopt vanilla contrastive learning technologies proposed for high-level visual tasks to low-level image restoration problems straightly. Because the acquired high-level global visual representations are insufficient for low-level tasks requiring rich texture and context information. In this article, we investigate the contrastive learning-based single-image super-resolution (SISR) from two perspectives: positive and negative sample construction and feature embedding. The existing methods take naive sample construction approaches (e.g., considering the low-quality input as a negative sample and the ground truth as a positive sample) and adopt a prior model (e.g., pretrained very deep convolutional networks proposed by visual geometry group (VGG) model) to obtain the feature embedding. To this end, we propose a practical contrastive learning framework for SISR (PCL-SR). We involve the generation of many informative positive and hard negative samples in frequency space. Instead of utilizing an additional pretrained network, we design a simple but effective embedding network inherited from the discriminator network, which is more task-friendly. Compared with the existing benchmark methods, we retrain them by our proposed PCL-SR framework and achieve superior performance. Extensive experiments have been conducted to show the effectiveness and technical contributions of our proposed PCL-SR thorough ablation studies. The code and resulting models will be released via https://github.com/Aitical/PCL-SISR.

Self-Supervised Arbitrary-Scale Implicit Point Clouds Upsampling.

Point clouds upsampling (PCU), which aims to generate dense and uniform point clouds from the captured sparse input of 3D sensor such as LiDAR, is a practical yet challenging task. It has potential applications in many real-world scenarios, such as autonomous driving, robotics, AR/VR, etc. Deep neural network based methods achieve remarkable success in PCU. However, most existing deep PCU methods either take the end-to-end supervised training, where large amounts of pairs of sparse input and dense ground-truth are required to serve as the supervision; or treat up-scaling of different factors as independent tasks, where multiple networks are required for different scaling factors, leading to significantly increased model complexity and training time. In this article, we propose a novel method that achieves self-supervised and magnification-flexible PCU simultaneously. No longer explicitly learning the mapping between sparse and dense point clouds, we formulate PCU as the task of seeking nearest projection points on the implicit surface for seed points. We then define two implicit neural functions to estimate projection direction and distance respectively, which can be trained by the pretext learning tasks. Moreover, the projection rectification strategy is tailored to remove outliers so as to keep the shape of object clear and sharp. Experimental results demonstrate that our self-supervised learning based scheme achieves competitive or even better performance than state-of-the-art supervised methods.

Application of deep learning algorithm in the recognition of cryptococcosis and talaromycosis skin lesions.

BACKGROUND: Cryptococcosis and talaromycosis are known as 'neglected epidemics' due to their high case fatality rates and low concern. Clinically, the skin lesions of the two fungal diseases are similar and easily misdiagnosed. Therefore, this study aims to develop an algorithm to identify cryptococcosis/talaromycosis skin lesions. METHODS: Skin images of tararomiasis and cryptococcosis were collected from published articles and augmented using the Python Imaging Library (PIL). Then, five deep artificial intelligence models, VGG19, MobileNet, InceptionV3, Incept ResNetV2 and DenseNet201, were developed based on the collected datasets using transfer learning technology. Finally, the performance of the models was evaluated using sensitivity, specificity, F1 score, accuracy, AUC and ROC curve. RESULTS: In total, 159 articles (79 for cryptococcosis and 80 for talaromycosis), including 101 cryptococcosis skin lesion images and 133 talaromycosis skin lesion images, were collected for further mode construction. Five methods showed good performance for prediction but did not yield satisfactory results for all cases. Among them, DenseNet201 performed best in the validation set, followed by InceptionV3. However, InceptionV3 showed the highest sensitivity, accuracy, F1 score and AUC values in the training set, followed by DenseNet201. The specificity of DenseNet201 in the training set is better than that of InceptionV3. CONCLUSIONS: DenseNet201 and InceptionV3 are equivalent to the optimal model in these conditions and can be used in clinical settings as decision support tools for the identification and classification of skin lesions of cryptococcus/talaromycosis.

Deep Attentional Guided Image Filtering.

Guided filter is a fundamental tool in computer vision and computer graphics, which aims to transfer structure information from the guide image to the target image. Most existing methods construct filter kernels from the guidance itself without considering the mutual dependency between the guidance and the target. However, since there typically exist significantly different edges in two images, simply transferring all structural information from the guide to the target would result in various artifacts. To cope with this problem, we propose an effective framework named deep attentional guided image filtering, the filtering process of which can fully integrate the complementary information contained in both images. Specifically, we propose an attentional kernel learning module to generate dual sets of filter kernels from the guidance and the target and then adaptively combine them by modeling the pixelwise dependency between the two images. Meanwhile, we propose a multiscale guided image filtering module to progressively generate the filtering result with the constructed kernels in a coarse-to-fine manner. Correspondingly, a multiscale fusion strategy is introduced to reuse the intermediate results in the coarse-to-fine process. Extensive experiments show that the proposed framework compares favorably with the state-of-the-art methods in a wide range of guided image filtering applications, such as guided super-resolution (SR), cross-modality restoration, and semantic segmentation. Moreover, our scheme achieved the first place in the real depth map SR challenge held in ACM ICMR 2021. The codes can be found at https://github.com/zhwzhong/DAGF.