Your browser doesn't support javascript.
loading
Mostrar: 20 | 50 | 100
Resultados 1 - 20 de 3.843
Filtrar
1.
Enzyme Microb Technol ; 149: 109850, 2021 Sep.
Artigo em Inglês | MEDLINE | ID: mdl-34311887

RESUMO

The rare sugar d-allulose is an attractive sucrose substitute due to its sweetness and ultra-low caloric value. It can be produced from D-fructose using d-allulose 3-epimerase (DAE) as the biocatalyst. However, most of the reported DAEs show low catalytic efficiency and poor thermostability, which limited their further use in food industrial. Here, a putative d-allulose 3-epimerase from a thermophilic organism of Halanaerobium congolense (HcDAE) was characterized, showing optimal activity at pH 8.0 and 70 °C in the presence of Mg2+. Saturation mutagenesis of Y7, C66, and I108, the putative residues responsible for substrate recognition at the O-4, -5, and -6 atoms of D-fructose was performed, and it yielded the triple mutant Y7H/C66L/I108A with improved activity toward D-fructose (345 % of wild-type enzyme). The combined mutant Y7H/C66L/I108A/R156C/K260C exhibited a half-half (t1/2) of 5.2 h at 70 °C and an increase of the Tm value by 6.5 °C due to the introduction of disulfide bridges between intersubunit with increased interface interactions. The results indicate that mutants could be used as industrial biocatalysts for d-allulose production.

2.
Int J Biol Macromol ; 186: 109-124, 2021 Jul 07.
Artigo em Inglês | MEDLINE | ID: mdl-34242645

RESUMO

Chemokines are crucial regulators of cell mobilization for development, homeostasis, and immunity. Chemokines signal through binding to chemokine receptors, a superfamily of seven-transmembrane domain G-coupled receptors. In the present study, seventeen CXC chemokine ligands (SsCXCLs) and nine CXC chemokine receptors (SsCXCRs) were systematically identified from Sebastes schlegelii genome. Phylogeny, synteny, and evolutionary analyses were performed to annotate these genes, indicating that the tandem duplications (CXCL8, CXCL11, CXCL32, CXCR2, and CXCR3), the whole genome duplications (CXCL8, CXCL12, CXCL18, and CXCR4), and the teleost-specific members (CXCL18, CXCL19, and CXCL32) led to the expansion of SsCXCLs and SsCXCRs. In addition, SsCXCLs and SsCXCRs were ubiquitously expressed in nine examined healthy tissues, with high expression levels observed in head kidney, liver, gill and spleen. Moreover, most SsCXCLs and SsCXCRs were significantly differentially expressed in head kidney, liver, and gill after Aeromonas salmonicida infection, and exhibited tissue-specific and time-dependent manner. Finally, protein-protein interaction network (PPI) analysis indicated that SsCXCLs and SsCXCRs interacted with a few immune-related genes such as interleukins, cathepsins, CD genes, and TLRs, etc. These results should be valuable for comparative immunological studies and provide insights for further functional characterization of chemokines and receptors in teleost.

3.
Molecules ; 26(14)2021 Jul 16.
Artigo em Inglês | MEDLINE | ID: mdl-34299586

RESUMO

Protein glycosylation is important in many organisms for proper protein folding, signaling, cell adhesion, protein-protein interactions, and immune responses. Thus, effectively determining the extent of glycosylation in glycoprotein therapeutics is crucial. Up to now, characterizing protein glycosylation has been carried out mostly by liquid chromatography mass spectrometry (LC-MS), which requires careful sample processing, e.g., glycan removal or protein digestion and glycopeptide enrichment. Herein, we introduce an NMR-based method to better characterize intact glycoproteins in natural abundance. This non-destructive method relies on exploiting differences in nuclear relaxation to suppress the NMR signals of the protein while maintaining glycan signals. Using RNase B Man5 and RNase B Man9, we establish reference spectra that can be used to determine the different glycoforms present in heterogeneously glycosylated commercial RNase B.

4.
Zhongguo Yi Xue Ke Xue Yuan Xue Bao ; 43(3): 435-444, 2021 Jun 30.
Artigo em Chinês | MEDLINE | ID: mdl-34238421

RESUMO

Circular RNA(circRNA)is a novel type of endogenous non-coding RNA.Most circRNAs act as microRNA(miRNA)sponges to regulate the expression of functional genes.In addition,some circRNAs can be translated and interact with RNA-binding proteins to perform biological functions.The expression of circRNAs is prevalent in tissues and body fluids,and their abnormal expression is related to tumor progression.circRNAs are stable even under the treatment of RNase R because of their circular conformation.As circRNAs have construct stability,wide variety,specific regulation of tumor progression and high expression in body fluids,it is potential for circRNAs to serve as candidate diagnostic,prognostic and therapeutic targets.However,the knowledge about circRNAs remains poor.In addition to the not completely resolved functions and generation mechanisms of circRNAs,the annotations of circRNAs are also waiting for expanding.Here,we review the generation mechanisms,biological functions,and application of circRNAs in tumor research,aiming to provide integrated information for the future research.


Assuntos
MicroRNAs , RNA Circular , Biomarcadores Tumorais/genética , Prognóstico
5.
Artigo em Inglês | MEDLINE | ID: mdl-34204145

RESUMO

Depressive symptoms are a common mental health problem among adolescents, which may affect their physical and mental health development and impose heavy burdens on individual families and society. This study aimed to examine the associations between sleep duration, academic pressure, and depressive symptoms among Chinese adolescents and to construct the mediation model to explore the mediating effect of sleep duration. The data are from the China Family Panel Studies (CFPS). Methodologically, the aforementioned associations were explored by constructing a structural equation model and applying multivariate multilevel logistic regression. In this study, we found that approximately 6.49% of the 3724 Chinese adolescents had depressive symptoms. Sleep duration of <6 h/night (OR = 2.39, 95%CI = 1.33-4.32) and high/maximum academic pressure (high: OR = 1.43, 95%CI = 1.02-1.99; maximum: OR = 2.43, 95%CI = 1.58-3.73) were both associated with an increased risk of depressive symptoms in adolescents. Meanwhile, the multiplicative interaction between sleep duration and academic pressure was significantly associated with depressive symptoms in adolescents (p < 0.001). The sleep duration played a partial mediating role in the relationship between academic pressure and depressive symptoms (a*b = 0.006, 95%BootCI = 0.001-0.012). Our study highlights that it is essential to mitigate the academic pressure of adolescents to increase their sleep duration and further reduce the occurrence of depressive symptoms by adopting corresponding preventive measures.


Assuntos
Depressão , Saúde Mental , Adolescente , Grupo com Ancestrais do Continente Asiático , China/epidemiologia , Depressão/epidemiologia , Humanos , Sono
7.
J Acoust Soc Am ; 149(6): 4596, 2021 Jun.
Artigo em Inglês | MEDLINE | ID: mdl-34241419

RESUMO

In this paper, a meta-learning-based underwater acoustic (UWA) orthogonal frequency division multiplexing (OFDM) system is proposed to deal with the environment mismatch in real-world UWA applications, which can effectively drive the model from the given UWA environment to the new UWA environment with a relatively small amount of data. With meta-learning, we consider multiple UWA environments as multi-UWA-tasks, wherein the meta-training strategy is utilized to learn a robust model from previously observed multi-UWA-tasks, and it can be quickly adapted to the unknown UWA environment with only a small number of updates. The experiments with the at-sea-measured WATERMARK dataset and the lake trial indicate that, compared with the traditional UWA-OFDM system and the conventional machine learning-based framework, the proposed method shows better bit error rate performance and stronger learning ability under various UWA scenarios.

8.
ACS Sens ; 2021 Jul 20.
Artigo em Inglês | MEDLINE | ID: mdl-34282892

RESUMO

Sensitive, selective, rapid, and label-free detection of pathogenic bacteria with high generality is of great importance for clinical diagnosis, biosecurity, and public health. However, most traditional approaches, such as microbial cultures, are time-consuming and laborious. To circumvent these problems, surface-enhanced Raman spectroscopy (SERS) appears to be a powerful technique to characterize bacteria at the single-cell level. Here, by SERS, we report a strategy for the rapid and specific detection of 22 strains of common pathogenic bacteria. A novel and high-quality silver nanorod SERS substrate, prepared by the facile interface self-assembly method, was utilized to acquire the chemical fingerprint information of pathogens with improved sensitivity. We also applied the mathematical analysis methods, such as the t-test and receiver operating characteristic method, to determine the Raman features of these 22 strains and demonstrate the clear identification of most bacteria (20 strains) from the rest and also the reliability of this SERS sensor. This rapid and specific strategy for wide-range bacterial detection offers significant advantages over existing approaches and sets the base for automated and onsite detection of pathogenic bacteria in a complex real-life situation.

9.
Neurocrit Care ; 2021 Jul 20.
Artigo em Inglês | MEDLINE | ID: mdl-34286462

RESUMO

BACKGROUND: Proliferation and apoptosis of vascular smooth muscle cells (VSMCs) are linked to intracranial aneurysm (IA) formation and progression. Long antisense noncoding RNA in the INK4 locus (ANRIL) has been reported to regulate VSMC functions in several cardiovascular diseases. However, little is known about how ANRIL influences VSMC proliferation and apoptosis during IA pathogenesis. METHODS: The expression level of ANRIL in the plasma and arterial wall tissues of patients with IA was detected by real-time quantitative polymerase chain reaction. The functional role of ANRIL in the regulation of VSMC proliferation and apoptosis and its downstream regulatory mechanism were determined using Cell Counting Kit 8, immunofluorescence, terminal-deoxynucleotidyl transferase-mediated UTP nick end labeling, western blotting, luciferase reporter assay, and RNA immunoprecipitation assay. RESULTS: ANRIL was downregulated in the plasma and arterial wall tissues of patients with IA, when compared with control groups. Overexpression of ANRIL significantly promoted VSMC proliferation and blocked cell apoptosis. Mechanistic studies demonstrated that ANRIL directly bound to microRNA-7 (miR-7) and that overexpression of miR-7 overturned the increased cell proliferation and decreased cell apoptosis, which was induced by ANRIL restoration. Besides, further study showed that ANRIL positively regulated fibroblast growth factor 2 (FGF2) expression via targeting miR-7. CONCLUSIONS: These results suggested that ANRIL affects VSMC proliferation and apoptosis via regulation of the miR-7/FGF2 pathway in IA, which provided a potential novel strategy for the treatment of IA.

10.
Int J Biol Macromol ; 2021 Jul 19.
Artigo em Inglês | MEDLINE | ID: mdl-34293363

RESUMO

The critical roles of transcription factors in cell differentiation and the delineation of cell phenotypes have been reported. The current study aimed to characterize the functions of the basic transcription factor 3 (BTF3) gene and its regulation of the intestinal stem cell marker B cell-specific Moloney murine leukemia virus insertion site 1 (BMI1) gene in colorectal cancer (CRC). Stem cell-like traits and epithelial-mesenchymal transition (EMT) of cultured human CRC cell line HCT116 were evaluated by CD133+ subpopulation counting, colony formation, tumorosphere generation, and expression of EMT-specific markers and stem cell markers. The interaction of BTF3 with BMI1 was analyzed. BTF3 was overexpressed in CRC tissues, which was associated with poor patient survival. BTF3 knockdown impaired the retention of stem cell-like traits of HCT116 and inhibited the EMT of HCT116 cells. BMI1 expression changed in a BTF3-dependent manner, and its overexpression could partially restore stem cell-like traits and EMT of cultured HCT116 cells after BTF3 knockdown. In parallel, treatment with the BMI1 inhibitor PTC-209 mimicked the effects of BTF3 knockdown on stem cell-like traits and EMT of cultured HCT116 cells. Together, these results support the notion that BTF3 and BMI1 are potential therapeutic targets to limit CRC metastasis.

11.
Sci Rep ; 11(1): 15040, 2021 Jul 22.
Artigo em Inglês | MEDLINE | ID: mdl-34294834

RESUMO

Colorectal cancer (CRC) ranks fourth among the deadliest cancers globally, and the progression is highly affected by the tumor microenvironment (TME). This study explores the relationship between TME and colorectal cancer prognosis and identifies prognostic genes related to the CRC microenvironment. We collected the gene expression data from The Cancer Genome Atlas (TCGA) and calculated the scores of stromal/immune cells and their relations to clinical outcomes in colorectal cancer by the ESTIMATE algorithm. Lower immune scores were significantly related to the malignant progression of CRC (metastasis, p = 0.001). We screened 292 differentially expressed genes (DEGs) by dividing CRC cases into high and low stromal/immune score groups. Functional enrichment analyses and protein-protein interaction (PPI) networks illustrated that these DEGs were closely involved in immune response, cytokine-cytokine receptor interaction, and chemokine signaling pathway. Six DEGs (FABP4, MEOX2, MMP12, ERMN, TNFAIP6, and CHST11) with prognostic value were identified by survival analysis and validated in two independent cohorts (GSE17538 and GSE161158). The six DEGs were significantly related to immune cell infiltration levels based on the Tumor Immune Estimation Resource (TIMER). The results might contribute to discovering new diagnostic and prognostic biomarkers and new treatment targets for colorectal cancer.

12.
Genes Cells ; 2021 Jul 06.
Artigo em Inglês | MEDLINE | ID: mdl-34228857

RESUMO

Bronchopulmonary dysplasia (BPD) is an inflammation-related respiratory disorder in infants. MiR-382-5p has displayed low expression in developing lungs with BPD, while the effect of miR-382-5p on BPD remains elusive. Here, a hyperoxia (85% oxygen)-induced BPD model in neonatal mice was established. On postnatal days 10 and 15, hyperoxia reduced miR-382-5p expression in lungs of mice. Besides, CDK8, CD68 and CD86 levels were elevated on day 15 after birth, implying the involvement of CDK8 in M1 macrophage polarization. In addition, in vitro injury in RAW264.7 macrophages was induced by IFN-γ and LPS stimulation. Lentivirus-encoding miR-382-5p decreased CDK8 expression, alleviated the production of inflammatory cytokines TNF-α, IL-1ß and IL-6, and restricted the levels of CD40 and CD86 in response to IFN-γ and LPS. Moreover, miR-382-5p inhibited the phosphorylation of STAT1. Luciferase reporter assay verified that miR-382-5p might target the 3'UTR of CDK8. Rescue assays revealed that CDK8 reversed the mitigating roles of miR-382-5p in inflammatory response and M1 macrophage polarization, as reflected by increased IL-6 and CD40 levels. Taken together, these findings indicate that miR-382-5p may suppress M1 macrophage activation and inflammatory response via inhibiting CDK8, thereby regulating the development of BPD, which is possibly mediated by STAT1 signaling.

13.
Chem Commun (Camb) ; 57(55): 6804-6807, 2021 Jul 08.
Artigo em Inglês | MEDLINE | ID: mdl-34236361

RESUMO

Glycosylation plays important roles in SARS-CoV-2 infection. We describe here a facile chemoenzymatic synthesis of core-fucosylated N-glycopeptides derived from the SARS-CoV-2 Spike protein and their binding with glycan-dependent neutralizing antibody S309 and human lectin CLEC4G. The synthetic glycopeptides provide tools for further functional characterization of viral glycosylation.


Assuntos
Glicopeptídeos/síntese química , Glicopeptídeos/metabolismo , Glicoproteína da Espícula de Coronavírus/química , Anticorpos Neutralizantes/imunologia , Técnicas de Química Sintética , Glicopeptídeos/química , Glicopeptídeos/imunologia , Glicosilação , Polissacarídeos/metabolismo
14.
Commun Biol ; 4(1): 882, 2021 Jul 16.
Artigo em Inglês | MEDLINE | ID: mdl-34272468

RESUMO

Cytosine or adenine base editors (CBEs or ABEs) hold great promise in therapeutic applications because they enable the precise conversion of targeted base changes without generating of double-strand breaks. However, both CBEs and ABEs induce substantial off-target DNA editing, and extensive off-target RNA single nucleotide variations in transfected cells. Therefore, the potential effects of deaminases induced by DNA base editors are of great importance for their clinical applicability. Here, the transcriptome-wide deaminase effects on gene expression and splicing is examined. Differentially expressed genes (DEGs) and differential alternative splicing (DAS) events, induced by base editors, are identified. Both CBEs and ABEs generated thousands of DEGs and hundreds of DAS events. For engineered CBEs or ABEs, base editor-induced variants had little effect on the elimination of DEGs and DAS events. Interestingly, more DEGs and DAS events are observed as a result of over expressions of cytosine and adenine deaminases. This study reveals a previously overlooked aspect of deaminase effects in transcriptome-wide gene expression and splicing, and underscores the need to fully characterize such effects of deaminase enzymes in base editor platforms.

15.
Sensors (Basel) ; 21(14)2021 Jul 12.
Artigo em Inglês | MEDLINE | ID: mdl-34300499

RESUMO

The compressive sensing (CS)-based sparse channel estimator is recognized as the most effective solution to the excessive pilot overhead in massive MIMO systems. However, due to the complex signal processing in the wireless communication systems, the measurement matrix in the CS-based channel estimation is sometimes "unfriendly" to the channel recovery. To overcome this problem, in this paper, the state-of-the-art sparse Bayesian learning using approximate message passing with unitary transformation (UTAMP-SBL), which is robust to various measurement matrices, is leveraged to address the multi-user uplink channel estimation for hybrid architecture millimeter wave massive MIMO systems. Specifically, the sparsity of channels in the angular domain is exploited to reduce the pilot overhead. Simulation results demonstrate that the UTAMP-SBL is able to achieve effective performance improvement than other competitors with low pilot overhead.

16.
Schizophr Res ; 2021 Jul 06.
Artigo em Inglês | MEDLINE | ID: mdl-34244046

RESUMO

BACKGROUND: Increasing evidence suggests that major psychiatric disorders, including major depressive disorder (MDD), bipolar disorder (BD) and schizophrenia (SZ) share biological, neuropsychological and clinical features, despite the criteria for their respective diagnoses being different. Neuroimaging studies have shown disrupted 'static' neural connectivity in these disorders. However, the changes in brain dynamics across the three psychiatric disorders remain unknown. METHODS: We aim to examine the connections and divergencies of the dynamic amplitude of low-frequency fluctuation (dALFF) in MDD, BD and SZ. In total, 901 participants [MDD, 229; BD, 146; SZ, 142; and healthy controls (HCs), 384] received resting-state functional magnetic resonance imaging. The dALFF was calculated using sliding-window analysis and compared across three psychiatric disorders. RESULTS: We found significant increases of dALFF in the right fusiform, right hippocampus, right parahippocampal in participants with MDD, BD and SZ compared to HC. We also found specific increased dALFF changes in caudate and left thalamus for SZ and BD and decreased dALFF changes in calcarine and lingual for SZ and MDD. CONCLUSION: Our study found significant intrinsic brain activity changes in the limbic system and primary visual area in MDD, BD, and SZ, which indicates these areas disruptions are core neurobiological features shared among three psychiatric disorders. Meanwhile, our findings also indicate that specific alterations in MDD, BD, and SZ provide neuroimaging evidence for the differential diagnosis of the three mental disorders.

17.
Clin Exp Rheumatol ; 2021 Jun 22.
Artigo em Inglês | MEDLINE | ID: mdl-34251318

RESUMO

OBJECTIVES: Juvenile dermatomyositis (JDM) is an autoimmune disease characterised by a great heterogeneity in its clinical manifestations. In this study, we aimed to investigate the association between different clinical subtypes, laboratory data, and myositis antibodies of JDM. METHODS: A total of 132 JDM patients were enrolled and their medical records were retrospectively reviewed and autoantibodies tested. Twenty-one variables, including clinical manifestations and laboratory findings, were selected for analysis. We selected principal component analysis (PCA) as a pre-processing method for cluster analysis to convert the 21 original variables into independent principal components. We then conducted a PCA-based cluster analysis in order to analyse the association between patient clusters and the clinical data, laboratory data, and myositis autoantibodies. RESULTS: We identified 4 distinct JDM subgroups by PCA-based cluster analysis, namely: cluster A, JDM patients with arthralgia and intense inflammation; cluster B, JDM patients with clinical manifestations of vasculitis; cluster C, hypermyopathic JDM patients; and cluster D, JDM patients with skin involvement. There were significant differences between the 4 groups in serum alkaline phosphatase levels, usage of aggressive immunosuppressive therapy, and autoantibody expression of anti-mi2, anti-MDA5, anti-Jo1, and anti-PM-Scl100. CONCLUSIONS: We conducted cluster analysis of a cohort of JDM patients and identified 4 subgroups that represented diverse characteristics in the distribution of laboratory data and myositis autoantibodies, indicating that multidimensional assessment of clinical manifestations is highly valuable and urgently needed in JDM patients. These subgroups may contribute to individualised treatments and improved JDM patient prognosis.

18.
Artigo em Inglês | MEDLINE | ID: mdl-34272578

RESUMO

Gut microbiota have a significant impact on host physiology and health, and host genetics and diet are considered as two important factors, but it is difficult to discriminate the influence of each single factor (host or diet) on gut microbiota under natural conditions. Moreover, current studies of avian microbiota mainly focus on domestic or captive birds, and it is still uncertain how host and diet take part in changing avian gut microbiota composition, diversity, and function in the wild. Here, high-throughput sequencing of 16S rRNA was used to identify the gut microbiota communities for sympatric wintering Great Bustards and Common Cranes at different diets. The results showed that 8.87% operational taxonomic units (OTUs) were shared among all sampling birds; in contrast, 39.43% of Kyoto Encyclopedia of Genes and Genomes (KEGG) functional pathways were common among all individuals, indicating the existence of gut microbiota conservatism both in microbiota structure and function. Microbiota abundance and diversity differed between Great Bustards and Common Cranes in a specific wintering site, and microbiota variation was detected for the same host species under two different sites, suggesting that the change of gut microbiota was induced by both host and diet. Furthermore, we found that changes of both microbial communities and functional pathways were larger between hosts than those between diets, which revealed that host might be the dominant factor determining microbiota characteristics and function, while diet further drove the divergence of gut microbiota. Gut microbiota functions appeared to be more conserved than bacterial community structure, indicating that different bacteria may function in a similar way, while microbiota OTU diversity might not be necessarily associated with functional diversity. With diet shifting, gut microbiota changed both in terms of microbial communities and functional pathways for the sympatric birds, which implies that avian habitats and their physiological microbiota would be influenced by different farmland management regimes. KEY POINTS: • Gut microbiota can be shaped by both diets and hosts in sympatric species. • Host was the dominant factor shaping the gut microbiota communities and functional pathways. • Gut microbiota were conservative both in structure and in function, but more conservative in function.

19.
J Colloid Interface Sci ; 604: 131-140, 2021 Jul 01.
Artigo em Inglês | MEDLINE | ID: mdl-34271486

RESUMO

Gold nanoparticles (Au NPs) with surface plasmonic resonance (SPR) effect and excellent internal electron transfer ability have widely been combined with semiconductors for photocatalysis. However, the in-depth effects of Au NPs in multicomponent photocatalysts have not been completely understood. Herein, ternary titanium oxide-gold-cadmium sulfide (TiO2-Au-CdS, TAC) photocatalysts, based on hierarchical TiO2 inverse opal photonic crystal structure with different Au NPs sizes have been designed to reveal the SPR effect and internal electron transfer of Au NPs in the presence of slow photon effect. It appears that the SPR effect and internal electron transfer ability of Au NPs, depending on their sizes, play a synergistic effect on the photocatalytic enhancement. The ternary TAC-10 photocatalyst with ~ 10 nm Au NPs demonstrates an unprecedented hydrogen evolution rate of 47.6 mmolh-1g-1 under visible-light, demonstrating ~ 48% enhancement comparing to the sample without slow photon effect. In particular, a 9.83% apparent quantum yield under 450 nm monochromatic light is achieved for TAC-10. A model is proposed and finite-difference time-domain (FDTD) simulations reveal the size influence of Au NPs in ternary TAC photocatalysts. This work suggests that the rational design of bifunctional Au NPs coupling with slow photon effect could largely promote hydrogen production from visible-light driven water splitting.

20.
Med Image Anal ; 73: 102148, 2021 Jul 01.
Artigo em Inglês | MEDLINE | ID: mdl-34274693

RESUMO

Deep learning models achieve strong performance for radiology image classification, but their practical application is bottle-necked by the need for large labeled training datasets. Semi-supervised learning (SSL) approaches leverage small labeled datasets alongside larger unlabeled datasets and offer potential for reducing labeling cost. In this work, we introduce NoTeacher, a novel consistency-based SSL framework which incorporates probabilistic graphical models. Unlike Mean Teacher which maintains a teacher network updated via a temporal ensemble, NoTeacher employs two independent networks, thereby eliminating the need for a teacher network. We demonstrate how NoTeacher can be customized to handle a range of challenges in radiology image classification. Specifically, we describe adaptations for scenarios with 2D and 3D inputs, uni and multi-label classification, and class distribution mismatch between labeled and unlabeled portions of the training data. In realistic empirical evaluations on three public benchmark datasets spanning the workhorse modalities of radiology (X-Ray, CT, MRI), we show that NoTeacher achieves over 90-95% of the fully supervised AUROC with less than 5-15% labeling budget. Further, NoTeacher outperforms established SSL methods with minimal hyperparameter tuning, and has implications as a principled and practical option for semi-supervised learning in radiology applications.

SELEÇÃO DE REFERÊNCIAS
DETALHE DA PESQUISA
...