Search

BACKGROUND: Macrophages are critical players in regulating innate and adaptive immunity in the tumor microenvironment (TME). The prognostic value of macrophages and their heterogeneous phenotypes in non-small cell lung cancer (NSCLC) is still uncertain. METHODS: Surgically-resected samples of 681 NSCLC cases were stained by multiplex immunofluorescence to examine macrophage phenotypes as well as the expression levels of program death-ligand 1 (PD-L1) on them in both tumor nest and tumor stroma, including pan-macrophage (CD68+), M1 (CD68+CD163−), and M2 macrophages (CD68+CD163+). Various other immune cell markers, including CD4, CD8, CD20, CD38, CD66B, FOXP3, and CD133, were also evaluated. Machine learning algorithm by Random Forest (RF) model was utilized to screen the robust prognostic markers and construct the CD68-based immune-related risk score (IRRS) for predicting disease-free survival (DFS). RESULTS: The expression levels of CD68 were moderately correlated with the levels of PD-L1 (P

BACKGROUND: Non-small cell lung cancer (NSCLC) patients with synchronous solitary metastasis are a heterogeneous population. The analysis and evaluation of NSCLC patients with synchronous solitary bone metastases by cTN stage (thoracic tumor staging) and volume parameters have not yet been studied. The purpose of this study is to estimate the prognostic value of cTN stage and volume parameters obtained by fluorodeoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) in NSCLC patients with synchronous solitary bone metastasis. METHODS: A total of 157 NSCLC patients with synchronous solitary bone metastasis were retrospectively analyzed. Patients’ cTN stage, metabolic tumor volume (MTV) parameters, and clinical data were collected. Kaplan-Meier survival analysis and a Cox regression model were performed to determine the association between each factor and overall survival (OS). Finally, time-dependent receiver operating characteristic (TDROC) curve analysis was used to assess the predictive capacity of the independent prognostic factors. RESULTS: Kaplan-Meier survival analysis showed significant differences between subgroups in terms of cTN stage. The median OS of group I was 44 months, and the 5-year survival rate was 39.6%. In the multivariate Cox regression analysis, cTN stage, MTV of the whole body (MTVwb), and MTV of thorax (MTVtho) were significantly associated with patient OS, even after adjusting for other clinical factors. However, MTV of bone (MTVbon) was not found to be an independent prognostic factor. TDROC curve analysis showed that cTN stage, MTVwb, and MTVtho had good predictive capacity for NSCLC patients with synchronous solitary bone metastasis. Compared with cTN stage and MTVtho, MTVwb had obviously better predictive specificity and sensitivity for the 5-year survival rate [5-year area under the curve (AUC) of MTVwb =0.844 vs. cTN stage (P=0.035) vs. MTVtho (P=0.052)]. The best cutoff value of MTVwb was 33.05. CONCLUSIONS: The results of this study confirmed that cTN stage, MTVwb, and MTVtho were independent prognostic factors of NSCLC patients with synchronous solitary bone metastases. These factors can be used for risk stratification of these patients. TDROC curve analysis indicated that cTN stage, MTVtho, and MTVwb had good performance for survival prediction.

The emergence of pretreatment drug resistance (PDR) caused by increased usage of antiretroviral therapy (ART) represents a significant challenge to HIV management. In this study, we evaluated the prevalence of PDR in people living with HIV (PLWH) in Chongqing, China.We retrospectively collected the data of 1110 ART-naïve PLWH in Chongqing from January 1, 2018 to June 30, 2021. HIV-1 genotypes and drug resistance were analyzed using the HIV-1 pol sequence. Risk factors associated with PDR were evaluated via the logistic regression model.Nine genotypes were detected among 1110 participants, with CRF07_BC (55.68%) being the dominant genotype, followed by CRF01_AE (21.44%), CRF08_BC (14.14%), and other genotypes (8.74%). Of all the participants, 24.14% exhibited drug resistance mutations (DRMs). The predominant DRMs for non-nucleoside reverse transcriptase inhibitors (NNRTIs) and nucleoside reverse transcriptase inhibitors (NRTIs) were V179D/E/A/DIN (13.60%) and M184V/I (1.44%), respectively, whereas only two major DRMs (M46L and I54L) were identified for protease inhibitors (PIs). The total prevalence of PDR was 10.54%, with 2.43%, 7.66%, and 1.71% participants exhibiting PDR to NRTIs, NNRTIs, and PIs, respectively. Furthermore, female PLWH, delays in ART initiation, and the CRF08_BC genotype were associated with a higher risk of PDR.Our study provides the first large cohort data on the prevalence of PDR in Chongqing, China. HIV-1 genotypes are diverse and complex, with a moderate level of PDR, which does not reach the threshold for the initiation of a public health response. Nevertheless, continuous surveillance of PDR is both useful and advisable.

Cellulose nanocrystals (CNC) were first isolated from microcrystalline cellulose (MCC) by p-toluene sulfonic acid (p-TsOH) hydrolysis. Cellulose II nanocrystal (CNC II) and cellulose III nanocrystal (CNC III) were then formed by swelling the obtained cellulose I nanocrystal (CNC I) in concentrated sodium hydroxide solutions and ethylenediamine (EDA) respectively. The properties of CNC I, CNC II and CNC III were subjected to comprehensive characterization by Fourier-transform infrared spectroscopy (FTIR), X-ray diffraction (XRD), scanning electron microscope (SEM), transmission electron microscopy (TEM), and thermogravimetric analysis (TGA). The results indicated that CNC I, CNC II and CNC III obtained in this research had high crystallinity index and good thermal stability. The degradation temperatures of the resulted CNC I, CNC II and CNC III were 300 ℃, 275 ℃ and 242 ℃, respectively. No ester bonds were found in the resulted CNC. CNC prepared in this research also had large aspect ratio and high negative zeta potential.

Cytopathology report generation is a necessary step for the standardized examination of pathology images. However, manually writing detailed reports brings heavy workloads for pathologists. To improve efficiency, some existing works have studied automatic generation of cytopathology reports, mainly by applying image caption generation frameworks with visual encoders originally proposed for natural images. A common weakness of these works is that they do not explicitly model the structural information among cells, which is a key feature of pathology images and provides significant information for making diagnoses. In this paper, we propose a novel graph-based framework called GNNFormer, which seamlessly integrates graph neural network (GNN) and Transformer into the same framework, for cytopathology report generation. To the best of our knowledge, GNNFormer is the first report generation method that explicitly models the structural information among cells in pathology images. It also effectively fuses structural information among cells, fine-grained morphology features of cells and background features to generate high-quality reports. Experimental results on the NMI-WSI dataset show that GNNFormer can outperform other state-of-the-art baselines., Comment: 12 pages, 6 figures

Graph data is ubiquitous in real world applications, as the relationship among entities in the applications can be naturally captured by the graph model. Finding cohesive subgraphs is a fundamental problem in graph mining with diverse applications. Given the important roles of cohesive subgraphs, this thesis focuses on cohesive subgraph identification in large graphs. Firstly, we study the size-bounded community search problem that aims to find a subgraph with the largest min-degree among all connected subgraphs that contain the query vertex q and have at least l and at most h vertices, where q, l, h are specified by the query. As the problem is NP-hard, we propose a branch-reduce-and-bound algorithm SC-BRB by developing nontrivial reducing techniques, upper bounding techniques, and branching techniques. Secondly, we formulate the notion of similar-biclique in bipartite graphs which is a special kind of biclique where all vertices from a designated side are similar to each other, and aim to enumerate all maximal similar-bicliques. We propose a backtracking algorithm MSBE to directly enumerate maximal similar-bicliques, and power it by vertex reduction and optimization techniques. In addition, we design a novel index structure to speed up a time-critical operation of MSBE, as well as to speed up vertex reduction. Efficient index construction algorithms are developed. Thirdly, we consider balanced cliques in signed graphs --- a clique is balanced if its vertex set can be partitioned into CL and CR such that all negative edges are between CL and CR --- and study the problem of maximum balanced clique computation. We propose techniques to transform the maximum balanced clique problem over G to a series of maximum dichromatic clique problems over small subgraphs of G. The transformation not only removes edge signs but also sparsifies the edge set.

Many seminal advances have been made in human immunodeficiency virus (HIV)/AIDS research over the past four decades. Treatment strategies, such as gene therapy and immunotherapy, are yielding promising results to effectively control HIV infection. Despite this, a cure for HIV/AIDS is not envisioned in the near future. A recently published academic study has raised awareness regarding a promising alternative therapeutic option for HIV/AIDS, referred to as “selective elimination of host cells capable of producing HIV” (SECH). Similar to the “shock and kill strategy,” the SECH approach requires the simultaneous administration of drugs targeting key mechanisms in specific cells to efficiently eliminate HIV replication-competent cellular reservoirs. Herein, we comprehensively review the specific mechanisms targeted by the SECH strategy. Briefly, the suggested cocktail of drugs should contain (i) latency reversal agents to promote the latency reversal process in replication-competent reservoir cells, (ii) pro-apoptotic and anti-autophagy drugs to induce death of infected cells through various pathways, and finally (iii) drugs that eliminate new cycles of infection by prevention of HIV attachment to host cells, and by HIV integrase inhibitor drugs. Finally, we discuss three major challenges that are likely to restrict the application of the SECH strategy in HIV/AIDS patients.

Background Patients with acquired immunodeficiency syndrome (AIDS) tend to suffer from several central nervous system (CNS) infections due to hypoimmunity. However, CNS aspergillosis (CNSAG) is extremely rare and difficult to diagnose. Thus, it is easily misdiagnosed. Case presentation We reported a 47-year-old male AIDS patient with ghosting vision and anhidrosis on the left head and face. He was accordingly diagnosed with Toxoplasma gondii encephalitis (TE) at other hospitals, for which he received regular anti-Toxoplasma gondii and anti-human immunodeficiency virus (anti-HIV) treatment. Then, the patient was transferred to our hospital due to a lack of any improvement with the prescribed treatment. The patient's neurological examination revealed no abnormalities at admission, only a slight change in the cerebrospinal fluid. His cranial magnetic resonance imaging (MRI) revealed multiple abnormal signals in the brain parenchyma, and his blood was positive for Toxoplasma gondii IgG antibody. The initial diagnosis at our hospital was also TE. Considering the poor efficacy of anti-TE treatment, cerebrospinal fluid metagenomics next-generation sequencing (mNGS) was performed, but no pathogenic bacteria were detected. However, Aspergillus fumigatus was detected in the cerebrospinal fluid via targeted next-generation sequencing (tNGS) and bronchoalveolar alveolar lavage fluid via mNGS. The diagnosis was accordingly revised to CNSAG combined with his other clinical manifestations. After administering voriconazole antifungal therapy, the patient’s symptoms were relieved, with improved absorption of the intracranial lesions. Conclusions The present case experience indicates the need for clinicians to strengthen their understanding of CNSAG. Moreover, for patients with diagnostic difficulties, early mNGS and tNGS (using biological samples with only a few pathogens) are helpful for early diagnosis and treatment, potentially allowing patients to achieve favorable outcomes.

Background The mortality rate remains high among patients with coinfection with Pneumocystis pneumonia (PCP) and HIV. The timing for initiation of antiretroviral therapy (ART) after a diagnosis of moderate to severe PCP remains controversial, however. We therefore designed the present study to determine the optimal timing for ART initiation in AIDS-associated PCP (AIDS/PCP) patients. Methods This was a multicenter, observational, prospective clinical trial. Eligible participants were recruited from 14 hospitals in mainland China, and assigned to an Early ART arm (initiation of ART ≤ 14 days after PCP diagnosis) and a Deferred ART arm (initiation of ART > 14 days after PCP diagnosis). The primary outcomes were death and the incidence of AIDS-defining events at week 48. The secondary outcomes were the changes in CD4+ T-cell counts from baseline values at weeks 12, 24, and 48, the virological suppression rate at week 24 and week 48, the rate of development of PCP-associated immune reconstitution inflammatory syndrome (PCP/IRIS), and the rate of adverse events over 48 weeks. Results The present study was performed using the data of 363 participants, with 169 participants in the Early ART arm, and 194 participants in the Deferred ART arm. Immunological and virological outcomes were found to be similar in both treatment arms. At week 48, there were no significant differences for the incidence of mortality (20 vs. 26, p = 0.860), and AIDS-defining events (17 vs. 26, p = 0.412). Over 48 weeks, the rates of PCP/IRIS (2 vs. 3, p = 1.000), adverse events (70 vs. 72, p = 0.465), and grade 3 or 4 adverse events (28 vs. 34, p = 0.919) did not reach statistical significance. A significant difference observed between two study arms was that 11 participants (55.0%) in the Early ART arm compared to 23 participants (88.5%) in the Deferred ART arm (p = 0.026) succumbed before ART had ever been started. Conclusions Early ART initiation results in no increase in mortality, AIDS-defining events, IRIS, adverse events, and immunological or virological outcomes. These results support the early initiation of ART in patients with moderate to severe AIDS/PCP. Clinical trial registration The present trial was registered at Chinese Clinical Trial Registry (ChiCTR1900021195). Registered 1 February 2019, http://www.chictr.org.cn/showproj.aspx?proj=35362.

A low-temperature adiabatic calorimeter was developed by Shandong Nonmetallic materials Institute in 2019, and the structure and mechanism were introduced briefly in this work. The heat capacities of the sample cells were measured and evaluated to improve the accuracy of the device. A high-purity benzene sample was selected as test sample to evaluate it molar heat capacity among the solid, solid–liquid interface and liquid state, and the results were generally consistent with the literature values. The melting point and molar heat enthalpy were calculated as 278.221 K and 9.814 kJ mol−1 based on the Raoult’s law, respectively. Measurement results were compared with literature data, and that indicated the adiabatic calorimeter possessed adequate accuracy in the thermodynamic properties measurement. A remarkable curvature of T − 1/F curve was appeared, and further researches were necessary to amend the purity calculation.

Background: Patients with acquired immunodeficiency syndrome (AIDS) tend to suffer from a complication of several central nervous system (CNS) infections owing to hypoimmunity. However, CNS aspergillosis (CNSAG) is extremely rare and difficult to diagnose, which makes it easily misdiagnosed. Case presentation: We have reported here a 47-year-old male AIDS patient with visual ghosting and no sweat development on the left head and face. He was accordingly diagnosed with toxoplasma gondii encephalitis (TE) in other hospitals, for which he received regular anti-toxoplasma gondii and anti-human immunodeficiency virus (anti-HIV) treatment. Then, the patient was transferred to our hospital due to lack of any improvement with the prescribed treatment. The patient's neurological examination revealed no abnormalities at admission, albeit only a slight change in the cerebrospinal fluid. His cranial magnetic resonance imaging (MRI) revealed multiple abnormal signals in the brain parenchyma, and his blood sample was positive for toxoplasma gondii IgG antibody. The initial diagnosis at our hospital was also TE. Considering the poor efficacy of anti-TE treatment, cerebrospinal fluid metagenomics next-generation sequencing (mNGS) was performed, but no pathogenic bacteria were detected. However, aspergillus fumigatus was detected in the cerebrospinal fluid with targeted next-generation sequencing (tNGS) and bronchofiberscope alveolar lavage fluid mNGS. The diagnosis was accordingly modified as CNSAG combined with his other clinical manifestations. After administering voriconazole anti-fungal therapy, the patient’s symptoms were relieved, with improved absorption of the intracranial lesions. Conclusions: The present case experience indicated the need for the clinicians to strengthen their understanding of CNSAG. Moreover, for patients with diagnostic difficulties, early mNGS and tNGS (using biological samples with only a few pathogens) were helpful for the early diagnosis and treatment, whereby the patients could achieve favorable outcomes.

Because of its poor engineering properties, silt often needs to be treated to meet the design requirements. Solidification using binder is one of the most common methods to improve the engineering properties of silt. Lime or/and cement are often used to treat silt. However, the mechanical properties of soil are closely related to its microstructure. Although much research has been done on the mechanical properties and stabilization mechanism of lime or/and cement solidified silt, the multi-scale research between the microstructure and macroscopic mechanics properties of solidified silt has received no attention. Therefore, in this paper, unconfined compression, mercury intrusion, and scanning electron microscope (SEM) tests were conducted on lime, cement, and lime and cement mixtures (LC) solidified silt. The microstructure parameters of particle, pore, and contact zone were extracted from SEM images. Meanwhile, the effect of curing time and binder dosage on the microstructure parameters was analyzed. In addition, the relationship between micro-parameters and macro-mechanical strength was established. The results showed that the particle diameter, particle area, contact zone diameter, and contact zone area increased with increasing curing time and binder dosage, while the pore ratio, fractal dimension of particle size distribution (PSD), and fractal dimension of contact zone size distribution (CZSD) decreased with increasing binder dosage. The unconfined compressive strength (UCS) decreased with the increase of particle roundness, pore area, pore diameter, pore ratio, and fractal dimension of PSD, and increased with increasing fractal dimension of pore size distribution (PPSD). In addition, there were no obvious relationships between UCS and pore roundness, contact zone area, and fractal dimension of CZSD.

We report acute antibody responses to SARS-CoV-2 in 285 patients with COVID-19. Within 19 days after symptom onset, 100% of patients tested positive for antiviral immunoglobulin-G (IgG). Seroconversion for IgG and IgM occurred simultaneously or sequentially. Both IgG and IgM titers plateaued within 6 days after seroconversion. Serological testing may be helpful for the diagnosis of suspected patients with negative RT-PCR results and for the identification of asymptomatic infections.

The deep-level traps induced by charged defects at the grain boundaries (GBs) of polycrystalline organic-inorganic halide perovskite (OIHP) films serve as major recombination centres, which limit the device performance. Herein, we incorporate specially designed poly(3-aminothiophenol)-coated gold (Au@PAT) nanoparticles into the perovskite absorber, in order to examine the influence of plasmonic resonance on carrier dynamics in perovskite solar cells. Local changes in the photophysical properties of the OIHP films reveal that plasmon excitation could fill trap sites at the GB region through photo-brightening, whereas transient absorption spectroscopy and density functional theory calculations correlate this photo-brightening of trap states with plasmon-induced interfacial processes. As a result, the device achieved the best efficiency of 22.0% with robust operational stability. Our work provides unambiguous evidence for plasmon-induced trap occupation in OIHP and reveals that plasmonic nanostructures may be one type of efficient additives to overcome the recombination losses in perovskite solar cells and thin-film solar cells in general.

Single-source domain generalization (SDG) in medical image segmentation is a challenging yet essential task as domain shifts are quite common among clinical image datasets. Previous attempts most conduct global-only/random augmentation. Their augmented samples are usually insufficient in diversity and informativeness, thus failing to cover the possible target domain distribution. In this paper, we rethink the data augmentation strategy for SDG in medical image segmentation. Motivated by the class-level representation invariance and style mutability of medical images, we hypothesize that unseen target data can be sampled from a linear combination of $C$ (the class number) random variables, where each variable follows a location-scale distribution at the class level. Accordingly, data augmented can be readily made by sampling the random variables through a general form. On the empirical front, we implement such strategy with constrained B$\acute{\rm e}$zier transformation on both global and local (i.e. class-level) regions, which can largely increase the augmentation diversity. A Saliency-balancing Fusion mechanism is further proposed to enrich the informativeness by engaging the gradient information, guiding augmentation with proper orientation and magnitude. As an important contribution, we prove theoretically that our proposed augmentation can lead to an upper bound of the generalization risk on the unseen target domain, thus confirming our hypothesis. Combining the two strategies, our Saliency-balancing Location-scale Augmentation (SLAug) exceeds the state-of-the-art works by a large margin in two challenging SDG tasks. Code is available at https://github.com/Kaiseem/SLAug .

The number of international benchmarking competitions is steadily increasing in various fields of machine learning (ML) research and practice. So far, however, little is known about the common practice as well as bottlenecks faced by the community in tackling the research questions posed. To shed light on the status quo of algorithm development in the specific field of biomedical imaging analysis, we designed an international survey that was issued to all participants of challenges conducted in conjunction with the IEEE ISBI 2021 and MICCAI 2021 conferences (80 competitions in total). The survey covered participants' expertise and working environments, their chosen strategies, as well as algorithm characteristics. A median of 72% challenge participants took part in the survey. According to our results, knowledge exchange was the primary incentive (70%) for participation, while the reception of prize money played only a minor role (16%). While a median of 80 working hours was spent on method development, a large portion of participants stated that they did not have enough time for method development (32%). 25% perceived the infrastructure to be a bottleneck. Overall, 94% of all solutions were deep learning-based. Of these, 84% were based on standard architectures. 43% of the respondents reported that the data samples (e.g., images) were too large to be processed at once. This was most commonly addressed by patch-based training (69%), downsampling (37%), and solving 3D analysis tasks as a series of 2D tasks. K-fold cross-validation on the training set was performed by only 37% of the participants and only 50% of the participants performed ensembling based on multiple identical models (61%) or heterogeneous models (39%). 48% of the respondents applied postprocessing steps.

Additional file 1: Fig. S1. Krona diagram of all species classification. Fig. S2. Heat map of relative abundance of 13 genera of HupA-producing endophytic fungi (a) in three tissues (R, root; S, stem; L, leaf) a (b) four seasons (M8, August; M11, November; M2, February; M5, May). Table S1. The negative correlation of HupA content with endophytic fungi within H. serrata (p < 0.05).

[spa] Los sistemas de visión artificial han surgido como alternativas competitivas en aplicacionesindustriales, siendo potenciadas actualmente en el camino hacia una industriaágil y flexible. Con su capacidad para lograr una alta precisión al tiempo que garantizaun alto rendimiento en la línea de producción, los sistemas de visión artificial tambiénhan ayudado a adoptar procesos de inspección y control de calidad de manera eficiente.En los últimos años, se ha informado que los sistemas de visión artificial que utilizanmetodologías de aprendizaje profundo (DL) pueden lograr un alto rendimiento, produciendodetecciones consistentes y precisas en diversas tareas de visión por computdormediante redes neuronales convolucionales profundas (DCNN).En esta tesis, abordamos los problemas de detección / reconocimiento de objetivosdesde dos puntos de vista diferentes: regresión de cajas circundantes (bounding boxes) ysegmentación semántica, ambos utilizando DCNN. Ambos enfoques se validan mediantedos aplicaciones conectadas con la industria, relacionadas con la inspección visual y elcontrol de calidad. Ambas soluciones intentan producir detecciones eficientes, ya sea pordiseño a través de la regresión de cajas circundantes rotadas, o por medio del etiquetadode píxeles individuales, la forma a través de la que la segmentación semántica se puedeadaptar, por naturaleza, a la forma de los objetos.En la primera parte de esta tesis, se propone una solución para el reconocimientode objetos en dos etapas. En este trabajo, se desarrolla una arquitectura piramidalbasada en el método Single-Shot multi-box Detector, con el objetivo de inferir cajas circundantesno rotadas. Posteriormente, desarrollamos una red de regresión sencilla parainferir las cajas rotadas sobre la base de las cajas no rotadas resultantes de la primeraetapa. En la segunda parte de esta tesis, desarrollamos soluciones de segmentación anivel de píxel que utilizan anotaciones completas y débiles. En cuanto a la solución totalmentesupervisada, entrenamos una red totalmente convolucional utilizando diferentesfunciones de pérdida, con el objetivo de resolver el problema de detección de objetivosde área pequeña. En cuanto al enfoque de segmentación semántica débilmente supervisada,proponemos una función de pérdida novedosa para contrarrestar los efectos delas anotaciones débiles. Por último, probamos varias estrategias simples para combinarla detección de cajas circundantes no orientados con enfoques de segmentación semánticacon el fin de obtener mejor rendimiento en comparación con la aplicación de losdos métodos de forma independiente. Todos los métodos mencionados anteriormente seevalúan utilizando datasets de las dos tareas de visión consideradas en esta tesis. Losresultados obtenidos nos permiten confirmar un rendimiento competitivo por parte delos métodos desarrollados. [eng] Machine vision systems have emerged as a superior alternative to human labor in industrialapplications, and are still being developed on the way to an agile and flexibleindustry. With its capability to achieve high accuracy while ensuring high throughputon the production line, machine vision systems have also helped to adopt inspectionand quality control processes efficiently. Over the past few years, machine vision systemsmaking use of Deep Learning (DL) methodologies have been reported to be ableto achieve high performance, producing consistent and accurate detections in variousvision tasks by means of Deep Convolutional Neural Networks (DCNN).In this dissertation we approach target detection/recognition problems from two differentpoints of view, namely bounding boxes regression and semantic segmentationusing DCNNs, and validate them by means of two industry-related applications connectedwith visual inspection and quality control. Both solutions try to produce efficientdetections, either by design through regression of rotated bounding boxes, or by meansof individual pixel labelling, the way how semantic segmentation adapts, by nature, tothe shape of the target.In the first part of this thesis, a two-stage solution for object recognition is proposed.In this work, a Feature Pyramid Architecture based on the Single Shot Multi-box Detectoris developed to infer unrotated bounding boxes. Subsequently, a lightweight regressionnetwork is designed to provide the rotated bounding boxes-based detections on thebasis of the resulting unrotated bounding boxes. In the second part of this thesis, pixellevelclassification solutions using full and weak annotations are developed. Regardingthe fully supervised solution, a fully convolutional network is trained using different lossfunctions, what aims at solving the detection problem for small area targets. As for theweakly supervised semantic segmentation approach, a novel loss function is proposedto counteract the effects of weak annotations. At last, we test several simple strategiesto combine the unoriented bounding boxes detection and semantic segmentationapproaches in order to get better performance compared with applying the two methodsindependently. All methods mentioned before are evaluated using datasets from the twovision tasks considered in this dissertation. The results obtained allow us to confirm thecompetitive performance achieved by the methods developed. [cat] Els sistemes de visió artificial han sorgit com alternatives competitives en aplicacions industrials,sent potenciades actualment en el camí cap a una indústria àgil i flexible. Ambla seva capacitat per aconseguir una alta precisió al temps que garanteix un alt rendimenten la línia de producció, els sistemes de visió artificial també han ajudat a adoptarprocessos d’inspecció i control de qualitat de manera eficient. En els últims anys, s’hainformat que els sistemes de visió artificial que utilitzen metodologies d’aprenentatge profund(DL) poden aconseguir un alt rendiment, produint deteccions consistents i precisesen diverses tasques de visió per computador mitjançant xarxes neuronals convolucionalsprofundes (DCNN).En aquesta tesi, abordem els problemes de detecció / reconeixement d’objectiusdes de dos punts de vista diferents: regressió de caixes circumdants (bounding boxes)i segmentació semàntica, ambdues utilitzant DCNN. Tots dos enfocaments es validenmitjançant dues aplicacions connectades amb la indústria, relacionades amb la inspeccióvisual i el control de qualitat. Les dues solucions intenten produir deteccions eficients,ja sigui per disseny a través de la regressió de caixes circumdants rotades, o per mitjà del’etiquetatge de píxels individuals, la forma mitjançant la qual la segmentació semànticaes pot adaptar, per naturalesa, a la forma dels objectes.A la primera part d’aquesta tesi, es proposa una solució per al reconeixement d’objectesen dues etapes. En aquest treball, es desenvolupa una arquitectura piramidal basada enel mètode Single-Shot multi-box Detector, amb l’objectiu d’inferir caixes circumdantsno rotades. Posteriorment, desenvolupam una xarxa de regressió senzilla per inferir lescaixes rotades sobre la base de les caixes no rotades resultants de la primera etapa. A lasegona part d’aquesta tesi, desenvolupem solucions de segmentació a nivell de píxel queutilitzen anotacions completes i febles. Pel que fa a la solució totalment supervisada, entrenemuna xarxa totalment convolucional utilitzant diferents funcions de pèrdua, ambl’objectiu de resoldre el problema de detecció d’objectius d’àrea petita. Pel que fa al’enfocament de segmentació semàntica feblement supervisada, proposem una funció depèrdua nova per contrarestar els efectes de les anotacions febles. Finalment, provamdiverses estratègies simples per combinar la detecció de caixes circumdants no orientadesamb enfocaments de segmentació semàntica per tal d’obtenir millor rendiment encomparació amb l’aplicació dels dos mètodes de forma independent. Tots els mètodesanteriorment esmentats s’avaluen utilitzant datasets de les dues tasques de visió consideradesen aquesta tesi. Els resultats obtinguts ens permeten confirmar un rendimentcompetitiu per part dels mètodes desenvolupats.

Image outpainting, which is well studied with Convolution Neural Network (CNN) based framework, has recently drawn more attention in computer vision. However, CNNs rely on inherent inductive biases to achieve effective sample learning, which may degrade the performance ceiling. In this paper, motivated by the flexible self-attention mechanism with minimal inductive biases in transformer architecture, we reframe the generalised image outpainting problem as a patch-wise sequence-to-sequence autoregression problem, enabling query-based image outpainting. Specifically, we propose a novel hybrid vision-transformer-based encoder-decoder framework, named \textbf{Query} \textbf{O}utpainting \textbf{TR}ansformer (\textbf{QueryOTR}), for extrapolating visual context all-side around a given image. Patch-wise mode's global modeling capacity allows us to extrapolate images from the attention mechanism's query standpoint. A novel Query Expansion Module (QEM) is designed to integrate information from the predicted queries based on the encoder's output, hence accelerating the convergence of the pure transformer even with a relatively small dataset. To further enhance connectivity between each patch, the proposed Patch Smoothing Module (PSM) re-allocates and averages the overlapped regions, thus providing seamless predicted images. We experimentally show that QueryOTR could generate visually appealing results smoothly and realistically against the state-of-the-art image outpainting approaches.

Despite the benefits achieved by the widespread availability of modern antiretroviral therapy (ART), HIV RNA integration into the host cell genome is responsible for the creation of latent HIV reservoirs, and represents a significant impediment to completely eliminating HIV infection in a patient via modern ART alone. Several methods to measure HIV reservoir size exist; however, simpler, cheaper, and faster tools are required in the quest for total HIV cure. Over the past few years, measurement of HIV-specific antibodies has evolved into a promising option for measuring HIV reservoir size, as they can be measured via simple, well-known techniques such as the western blot and enzyme-linked immunosorbent assay (ELISA). In this article, we re-visit the dynamic evolution of HIV-1-specific antibodies and the factors that may influence their levels in the circulation of HIV-positive individuals. Then, we describe the currently-known relationship between HIV-1-specific antibodies and HIV reservoir size based on study of data from contemporary literature published during the past 5 years. We conclude by highlighting current trends, and discussing the individual HIV-specific antibody that is likely to be the most reliable antibody for potential future utilization for quantification of HIV reservoir size.

We investigate the abnormal current-voltage (C-V) hump effect of p-type low-temperature polysilicon (LTPS) thin-film transistors (TFTs) which have undergone high current operations. Experimental results indicate localized electron trapping in the gate insulator (GI), which is carried out near the drain. The ON-current ( ${I}_{ \mathrm{\scriptscriptstyle ON}}$ ) enhancement is due to the reduction of effective length, and the OFF-current ( ${I}_{ \mathrm{\scriptscriptstyle OFF}}$ ) decrease as the electron tunneling path distance increases. These can be observed after hot carrier stress in current characteristics. The C-V measurements demonstrate that the threshold voltage ( ${V}_{\text {th}}$ ) shift is associated with the gate length. In addition, capacitance-voltage measurements also show that this localized trapping region remains the same in length, regardless of channel length. Hence, a model is proposed to explain how the electric field, which is gate length-dependent, affects the source side of the device, and then lowers the source barrier height. This leads to bulk leakage, which causes the subthreshold swing degradation at device scale down.

The wedge-tailed green pigeon (

Dynamical coupling between matter and gauge fields underlies the emergence of many exotic particle-like excitations in condensed matter and high energy physics. An important stepping stone to simulate this physics in atomic quantum gases relies on the synthesis of density-dependent gauge fields. Here we demonstrate deterministic formation of domain walls in a stable Bose-Einstein condensate with a synthetic gauge field that depends on the atomic density. The gauge field is created by simultaneous modulations of the optical lattice potential and interatomic interactions, and results in domains of atoms condensed into two different momenta. Modeling the domain walls as elementary excitations, we find that the domain walls respond to synthetic electric field with a charge-to-mass ratio larger than and opposite to that of the bare atoms. Our work offers promising prospects to simulate the dynamics and interactions of novel excitations in quantum systems with dynamical gauge fields.

The wedge-tailed green pigeon (Treron sphenurus) has a protective value in the evolution of the family Columbidae. In this study, the complete mitogenome of T. sphenurus from Baise City, China, which represents the first sequenced species of the genus Treron in Tribe Treronini, is reported. This was accomplished using PCR-based methods and a primer-walking sequencing strategy with genus-specific primers. The mitogenome was found to be 18,919 bp in length comprising 37 genes, including 13 protein-coding genes, two rRNA genes, 22 tRNA genes, and one control region. In terms of structure and composition, many similarities were found between the T. sphenurus and Hemiphaga novaeseelandiae (New Zealand pigeon) mitogenomes. This was further supported by phylogenetic analysis showing that T. sphenurus has a close evolutionary relationship with H. novaeseelandiae. The complete mitogenome of T. sphenurus reported here is expected to provide valuable molecular information for further studies on the phylogeny of the genus Treron and for analyses of the taxonomic status of the family Columbidae.

Automatic nuclei segmentation and classification play a vital role in digital pathology. However, previous works are mostly built on data with limited diversity and small sizes, making the results questionable or misleading in actual downstream tasks. In this paper, we aim to build a reliable and robust method capable of dealing with data from the 'the clinical wild'. Specifically, we study and design a new method to simultaneously detect, segment, and classify nuclei from Haematoxylin and Eosin (H&E) stained histopathology data, and evaluate our approach using the recent largest dataset: PanNuke. We address the detection and classification of each nuclei as a novel semantic keypoint estimation problem to determine the center point of each nuclei. Next, the corresponding class-agnostic masks for nuclei center points are obtained using dynamic instance segmentation. Meanwhile, we proposed a novel Joint Pyramid Fusion Module (JPFM) to model the cross-scale dependencies, thus enhancing the local feature for better nuclei detection and classification. By decoupling two simultaneous challenging tasks and taking advantage of JPFM, our method can benefit from class-aware detection and class-agnostic segmentation, thus leading to a significant performance boost. We demonstrate the superior performance of our proposed approach for nuclei segmentation and classification across 19 different tissue types, delivering new benchmark results., 12 pages,7 figures, journal

We consider unsupervised cell nuclei segmentation in this paper. Exploiting the recently-proposed unpaired image-to-image translation between cell nuclei images and randomly synthetic masks, existing approaches, e.g., CycleGAN, have achieved encouraging results. However, these methods usually take a two-stage pipeline and fail to learn end-to-end in cell nuclei images. More seriously, they could lead to the lossy transformation problem, i.e., the content inconsistency between the original images and the corresponding segmentation output. To address these limitations, we propose a novel end-to-end unsupervised framework called Aligned Disentangling Generative Adversarial Network (AD-GAN). Distinctively, AD-GAN introduces representation disentanglement to separate content representation (the underling spatial structure) from style representation (the rendering of the structure). With this framework, spatial structure can be preserved explicitly, enabling a significant reduction of macro-level lossy transformation. We also propose a novel training algorithm able to align the disentangled content in the latent space to reduce micro-level lossy transformation. Evaluations on real-world 2D and 3D datasets show that AD-GAN substantially outperforms the other comparison methods and the professional software both quantitatively and qualitatively. Specifically, the proposed AD-GAN leads to significant improvement over the current best unsupervised methods by an average 17.8% relatively (w.r.t. the metric DICE) on four cell nuclei datasets. As an unsupervised method, AD-GAN even performs competitive with the best supervised models, taking a further leap towards end-to-end unsupervised nuclei segmentation.

Objective To investigate the characteristics of hosts, antimicrobial susceptibility, and molecular epidemiology of mucoid serotype 3 Streptococcus pneumoniae (S. pneumoniae) isolated from children in China. Method S. pneumoniae isolates collected between January 2016 and December 2019 were analyzed. S. pneumoniae isolates with mucoid phenotype were selected visually, and serotype 3 isolates were confirmed by Quellung reaction. The antimicrobial susceptibility was measured by E-test. Multilocus sequence typing was used for clonal analysis. Results Twenty (3.04%) isolates of mucoid serotype 3 S. pneumoniae were identified from 657 clinical isolates, and all of them were noninvasive strains. The mean age of the hosts was 5.69 ± 3.28 years. The isolates included: 50.0% from the dissected tonsil or adenoid tissue in children with obstructive sleep apnea-hypopnea syndrome, 45.0% from sputum or bronchial lavages in children with pneumonia, and 5.0% from vaginal secretions of one patient with vulvovaginitis. All isolates were susceptible to penicillin, cefuroxime, ceftriaxone, meropenem, vancomycin, levofloxacin, trimethoprim/sulfamethoxazole, and rifampin but resistant to erythromycin. Sequence type (ST)505 and its clonal complex (CC) were the main genotypes (95%). Antimicrobial susceptibility of ST180 and ST505 were compared, and the minimum inhibitory concentration (MIC) of ST505 isolates was significantly higher than that of ST180 for tetracycline, trimethoprim/sulfamethoxazole, and meropenem. Conclusions Mucoid serotype 3 Streptococcus pneumoniae can be isolated from various body parts, among which the respiratory system is the most common. It can cause noninvasive infection in children, and it has high susceptibility to a variety of antibiotics, especially β-lactams, but is resistant to macrolides. CC505 is the novel clonal complex found in China, which may be related to the worldwide mainstream clonal complex (CC180) but has its own biological characteristics.

The prevalence of asymptomatic cryptococcal antigenemia (ACA) in human immunodeficiency virus (HIV) infected individuals has been observed to be elevated. The prevalence of ACA ranges from 1.3% to 13%, with different rates of prevalence in various regions of the world. We reviewed studies conducted internationally, and also referred to two established expert consensus guideline documents published in China, and we have concluded that Chinese HIV-infected patients should undergo cryptococcal antigen screening when CD4+ T-cell counts fall below 200 cells/μL and that the recommended treatment regimen for these patients follow current World Health Organization guidelines, although it is likely that this recommendation may change in the future. Early screening and optimized preemptive treatment for ACA is likely to help decrease the incidence of cryptococcosis, and is lifesaving. Further studies are warranted to explore issues related to the optimal management of ACA.

Machine vision systems are emerging as increasingly popular solutions in automated quality and process control applications. By allowing non-contact, and therefore non-destructive inspection, techniques based on machine vision are especially suitable when correct handling of the object under inspection is critical. This is the quality control problem that we address in this work: it consists of the detection of a series of control elements that are placed in boxes and bags that contain the surgical instruments with which surgeons and nurses must be supplied before starting a surgical procedure. The presence of these elements shows that the instruments have been correctly subjected to the required sanitation processes. To address this problem, in this work we detect rectangular areas of the image that contain the objects of interest (traceability label, seal, various paper-tape filters that change with temperature, etc.). Additionally, we consider the problem of detecting oriented rectangular areas (not parallel to the image axes) to improve localization. The detector makes use of deep convolutional neural networks that determine by regression the parameters of the rectangular areas, adopting a two-stage strategy: the first stage locates the object through a rectangular box (bounding box) parallel to the image axes, the second stage estimates the parameters of the maximal oriented box contained in the first and containing the object. After training the system with a collection of representative images, where the objects to be detected appear in various orientations and scales, the ability of the system to detect the control elements in a collection of test images has been verified. The experimentation has started with a previous clustering process to determine a set of scales and appropriate aspect ratios as starting points for the regressions. Next, various box parameterizations have been considered to determine which one allowed for the highest correct detection rate. The final results point out, for the non-oriented boxes, to an average precision (P) greater than 98% for an average sensitivity (R) greater than 93%, using 4 clusters; for the oriented boxes case, P = 66% and R = 73%, using a 2-parameter model to orient the boxes., This work is also supported by projects EU-H2020 BUGWRIGHT2 (GA 871260), PGC2018-095709-B-C21 (MCIU/AEI/FEDER, UE), and PROCOE/4/2017 (Govern Balear, 50% P.O. FEDER 2014-2020 Illes Balears).

The SC200 proton therapy system commissioned by the Hefei CAS Ion Medical and Technical Devices Co., Ltd. (HFCIM; Hefei, China) and the Joint Institute for Nuclear Research (JINR; Dubna, Russia) has made significant progress. A main radial beam diagnostic system (MRBDS) equipped with a new type of train probe was developed to satisfy the requirements of beam diagnosis. In this paper, the detailed design of the mechanical structure and electronics system of the MRBDS is presented. The electronics system, which includes hardware and software components, was tested and calibrated. The results show that measurement errors can be significantly reduced by the designed calibration procedures. The repeatability of the mechanical structure was also verified, and the experimental results indicate that the unidirectional repeatability of the positioning is better than 0.3 mm. Finally, the MRBDS was used to measure the phase lag of the particles in the beam, and the results showed a high degree of agreement with theoretical calculations, which proved the applicability and high efficiency of the MRBDS.

Background: Currently, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread globally, causing an unprecedented pandemic. However, there is no specific antiviral therapy for coronavirus disease 2019 (COVID-19). We conducted a clinical trial to compare the effectiveness of three antiviral treatment regimens in patients with mild to moderate COVID-19. Methods: This was a single-center, randomized, open-labeled, prospective clinical trial. Eligible patients with mild to moderate COVID-19 were randomized into three groups: ribavirin (RBV) plus interferon-α (IFN-α), lopinavir/ritonavir (LPV/r) plus IFN-α, and RBV plus LPV/r plus IFN-α at a 1:1:1 ratio. Each patient was invited to participate in a 28-d follow-up after initiation of an antiviral regimen. The outcomes include the difference in median interval to SARS-CoV-2 nucleic acid negativity, the proportion of patients with SARS-CoV-2 nucleic acid negativity at day 14, the mortality at day 28, the proportion of patients re-classified as severe cases, and adverse events during the study period. Results: In total, we enrolled 101 patients in this study. Baseline clinical and laboratory characteristics of patients were comparable among the three groups. In the analysis of intention-to-treat data, the median interval from baseline to SARS-CoV-2 nucleic acid negativity was 12 d in the LPV/r+IFN-α-treated group, as compared with 13 and 15 d in the RBV+IFN-α-treated group and in the RBV+LPV/r+ IFN-α-treated group, respectively (p=0.23). The proportion of patients with SARS-CoV-2 nucleic acid negativity in the LPV/r+IFN-α-treated group (61.1%) was higher than the RBV+ IFN-α-treated group (51.5%) and the RBV+LPV/r+IFN-α-treated group (46.9%) at day 14; however, the difference between these groups was calculated to be statistically insignificant. The RBV+LPV/r+IFN-α-treated group developed a significantly higher incidence of gastrointestinal adverse events than the LPV/r+ IFN-α-treated group and the RBV+ IFN-α-treated group. Conclusions: Our results indicate that there are no significant differences among the three regimens in terms of antiviral effectiveness in patients with mild to moderate COVID-19. Furthermore, the combination of RBV and LPV/r is associated with a significant increase in gastrointestinal adverse events, suggesting that RBV and LPV/r should not be co-administered to COVID-19 patients simultaneously. Clinical Trial Registration: www.ClinicalTrials.gov, ID: ChiCTR2000029387. Registered on January 28, 2019.

Health effects resulting from the smoke of carbonyl compounds (aldehydes and ketones) and metal-containing incense particles at temples during incense burning periods were evaluated at temple A (without incense reduction activities) and B (with incense reduction activities), Nantou County, in 2018. The predominant size fractions of particles were PM1, PM1–2.5, and PM2.5–10 at both temples. The total particle mass at temple A was approximately 1.1 times that of temple B due to incense reduction at temple B. The most abundant metal elements in all particle size fractions at both temples were Fe, Al, and Zn. Metal species of incense smoke are divided into three groups by hierarchical cluster analysis and heatmaps, showing higher metal contents in groups PM1, PM18–10, and PM18–2.5 at temple A. In contrast, higher metal contents were observed in PM18–10 and PM2.5–1 at temple B. Most of the carbonyl species were formaldehyde and acetaldehyde, released during incense burning periods, with concentrations ranging from 6.20 to 13.05 μg/m3 at both temples. The total deposited fluxes of particle-bound metals at temples A and B were determined to be 83.00% and 84.82% using the International Commission on Radiological Protection (ICRP) model. Health-risk assessments revealed that the risk values of metals and carbonyls were above recommended guidelines (10−6) at temple A. Since worshippers and staff are exposed to incense burning environments with poor ventilation over a long period, these toxic organic compounds and metals increase health risks in the respiratory tract. Therefore, incense reduction is important to achieve healthy temple environments.