This study aimed to screen high extracellular polysaccharides (EPS) producing lactic acid bacteria and systematically characterize the temporal metabolic dynamics underlying mixed-culture fermentation of Munage grape juice (MGJ). Lactobacillus acidophilus and Lactiplantibacillus plantarum were stepwise selected from ten commercial strains and co-fermented at a 1:1 ratio. Response surface methodology was employed to establish optimal fermentation parameters: an inoculum size of 2.5% (v/v), initial pH of 5.9, temperature of 37 °C, and fermentation duration of 32 h. Under these conditions, EPS production reached 1.26 ± 0.02 g/L. Untargeted metabolomics analysis revealed distinct temporal metabolic reprogramming. During the early stage (0-16 h), pyruvate metabolism pathways were significantly enriched, resulting in the rapid accumulation of organic acids (lactate, citrate), which drove a decline in pH and suppressed undesirable microorganisms. In the mid-stage (16-32 h), phenylpropanoid biosynthesis, glutathione metabolism, and related pathways were cooperatively activated, leading to markedly increased levels of phenolic acids (coumaric acid, isoeugenol), flavonoids, and amino acid derivatives (N-acetylglutamate, His-Trp-Phe), thereby conferring enhanced antioxidant activity and flavor precursors. Metabolic activity plateaued during the late stage (32-40 h) with few differential metabolites observed, confirming 32 h as the optimal fermentation endpoint. This study provides the first comprehensive temporal metabolic atlas of mixed LAB fermentation in MGJ, offering both technical parameters and theoretical support for the development of functional fermented beverages.
The dice coefficient score (DSC) serves as a key indicator of segmentation accuracy in brain tumor analysis. However, current methods often face a trade-off between model compactness and segmentation precision, where lightweight models reduce accuracy and complex models increase computational cost. To address this issue, we propose ENAS-Net, an evolutionary UNet architecture that integrates a customized UNet framework with an evolutionary neural architecture search (ENAS) strategy for optimized performance. The encoder employs ReLU activation, convolution, max-pooling, and dropout to extract hierarchical features, while the middle layers enhance representation through convolution and weighted aggregation. The decoder utilizes transposed convolution, concatenation, and dropout for accurate reconstruction of tumor regions. The evolutionary search systematically explores the supernet by varying encoder–decoder depth, filter sizes, activation functions, and dropout rates through crossover and mutation across generations. The optimized ENAS-Net employs Conv2D and Conv2DTranspose layers for efficient feature extraction and upsampling. Experimental results demonstrate that ENAS-Net achieves a DSC of 92.23, HD95 of 5.42 with 5.8 M parameters, outperforming existing models in both accuracy and efficiency. These results confirm ENAS-Net’s potential for real-time and resource-efficient brain tumor segmentation in clinical settings.
Accurate and efficient segmentation of heterogeneous brain tumors requires neural architectures that can simultaneously capture diverse tumor subregions with varying appearance, scale, and boundary characteristics while remaining computationally compact. However, manually designed networks with fixed architectures often struggle to balance expressive feature representation and model efficiency across such heterogeneity. To address this challenge, we propose EvoBTSeg, an evolutionary architecture search framework that automatically discovers compact and effective network configurations designed for heterogeneous brain tumor segmentation. A customized U-Net supernet is defined with parameterized encoder, bottleneck, and decoder components, enabling flexible control over network depth, filter size, activation function, and dropout rate. The encoder extracts hierarchical representations through convolution, ReLU activation, max-pooling, and dropout to accommodate heterogeneous tumor patterns, while the bottleneck refines global contextual features via convolution and weighted feature aggregation. The decoder progressively restores spatial resolution using transposed convolution and feature concatenation, preserving fine-grained boundary information critical for accurate delineation of small and diffuse tumor subregions. An evolutionary search strategy based on crossover and mutation is employed to explore the architectural search space in a data-driven manner, selecting architectures that balance contextual modeling and boundary precision. Experimental results on the BraTS 2021 dataset demonstrate that EvoBTSeg identifies a compact architecture with only 0.12 M parameters, achieving a dice similarity coefficient (DSC) of 92.83% and a Hausdorff distance (HD95) of 1.38 mm. Furthermore, EvoBTSeg demonstrates reliable generalization, maintaining competitive segmentation performance on BraTS 2020 (DSC 89.37%, HD95 1.33 mm) and the ACDC dataset (DSC 88.03%, HD95 1.41 mm). These results indicate that EvoBTSeg effectively discovers lightweight yet accurate architectures for heterogeneous brain tumor segmentation, highlighting its potential for practical clinical deployment.
Purpose: The incidence of ischemic stroke in young adults has been increasing. However, there is a lack of in-depth understanding of relevant clinical features, particularly in specific geographic and climatic settings. This study aimed to comprehensively investigate the clinical features, including etiology and characteristics, of ischemic stroke in young adults compared with elderly patients from an island population characterized by a subtropical monsoon climate. Patients and Methods: A total of 1,010 patients with ischemic stroke were included and divided into a young group (454 patients, aged 18-50 years) and an elderly group (556 patients, aged >50 years). Clinical and radiological data were collected and compared between the two groups. Continuous and categorical variables were compared using the t-test (or non-parametric tests) and chi-square test, respectively. Binary logistic regression analyses were used to investigate differences between the two groups. Results: Body mass index, leukocyte count, lymphocyte count, uric acid, and triglyceride levels were higher in the young group than in the elderly group, whereas vitamin B12 levels were lower. In addition, the proportions of cardiogenic embolism, other etiologies and unexplained strokes, and infarcts in the basal ganglia region were significantly higher in the young group. These results suggest agerelated differences in clinical characteristics within this island study population. Conclusion: High body mass index, leukocyte count, uric acid, triglyceride levels, and low vitamin B12 levels are associated with ischemic stroke in young adults. Young adult patients had a higher prevalence of basal ganglia infarction, with a different subtype distribution from elderly patients. These findings highlight the importance of considering geographic specificity and age-related differences when developing stroke prevention and management strategies, formulating public health policies, and allocating medical resources for ischemic stroke.
INTRODUCTION:Coronary Microvascular Dysfunction (CMVD) can lead to myocardial ischemia and increase the risk of adverse cardiovascular events. In clinical practice, early and accurate diagnosis of CMVD is essential for effective intervention and management. However, because CMVD and obstructive coronary artery disease share similar clinical presentations, distinguishing CMVD remains challenging. METHODS:We conducted a narrative review by searching PubMed, Web of Science, and Google Scholar. Two reviewers independently screened studies and extracted technical and clinical details from the included articles. RESULTS:This review elaborates on noninvasive tools for CMVD evaluation. Positron Emission Tomography (PET) quantifies Myocardial Blood Flow (MBF). Cardiovascular Magnetic Resonance (CMR) provides a quantitative myocardial perfusion reserve index. Myocardial computed tomography perfusion, in static or dynamic modes, enables concurrent anatomic and functional assessment. Echocardiography includes transthoracic Doppler-derived coronary flow reserve and myocardial contrast echocardiography for bedside perfusion evaluation. SPECT supports MBF quantification in selected settings. CMVD pathobiology is reflected in circulating biomarkers, with microRNAs showing promise. Artificial Intelligence (AI) and computational fluid dynamics can further assist in noninvasive CMVD diagnosis. DISCUSSION:Each imaging modality has distinct strengths and limitations. Blood biomarkers and computational models are promising for scalable clinical use, but require confirmation in prospective studies. CONCLUSION:Cardiovascular imaging and circulating biomarkers reveal CMVD-related changes in anatomy, hemodynamics, and metabolism, while computational models can improve diagnostic precision. Future research should include large, multicenter, prospective studies, such as trials comparing the diagnostic accuracy of AI-enhanced CMR with the invasive gold standard, the index of microvascular resistance, to validate these methods and establish integrated, cost-effective diagnostic pathways for early CMVD detection across diverse cohorts.
