Abstract A simple linear regression framework was developed to quantify contamination in pathogen metagenomic next‐generation sequencing (mNGS). The model was trained on simulated contamination samples generated by spiking HeLa cell backgrounds (1 × 10 5 cells/mL) with gradient concentrations of Streptococcus pneumoniae positive standards (5.6 × 10 6 copies/mL) to cover admixture ratios of 0.05%–10% (2.8 × 10 3 to 5.6 × 10 5 copies/mL). Samples were independently generated as two sets: a modeling set ( n = 178) and an independent validation set ( n = 32). Metagenomic libraries were prepared and sequenced, followed by quality trimming (fastp) and host sequence removal (Kraken2); non‐host normalized reads were calculated using a custom database based on the S. pneumoniae reference genome (NC_003028.3). After refinement via outlier exclusion and model rebuilding, the final model achieved an R 2 of 0.966 and an overall MAPE of 16.46% on an independent validation set, supporting clinically actionable contamination monitoring pending multi‐species validation.
Abstract Bessel beams offer superior depth of field and self‐healing properties compared to conventional Gaussian beams in light sheet fluorescence microscopy (LSFM), enabling enhanced axial resolution and optical sectioning. However, their side lobes generate significant out‐of‐focus fluorescence, which degrades image contrast and signal‐to‐noise ratio (SNR). In this context, we demonstrate the use of a physics‐guided, unsupervised deep learning model‐based image restoration pipeline for Bessel beam LSFM images to suppress the side‐lobe‐induced artifacts that does not require ground‐truth data. A synthetic point spread function (PSF) of the Bessel beam LSFM is simulated using the Debye‒Wolf model and wave‐optical simulations based on the optical parameters of the microscope. The modeled PSF is then used with a hybrid zero‐shot deconvolution and U‐Net model for denoising and contrast enhancement of fluorescent biological sample images without the requirement of paired ground‐truth data. The parameters of the deep learning model are optimized to obtain substantial image quality improvements in terms of SNR and contrast across different biological specimens, namely, mouse kidney tissues (in vitro) and three‐dimensional image stack of trabecular meshwork of a porcine eye sample (ex vivo). The results showed up to 2.2‐fold enhancement in SNR and 89% enhancement in Michelson contrast. The robustness of the method under moderate PSF mismatch is also investigated. The proposed approach holds significant potential in a variety of biomedical imaging applications and monitoring of sub‐cellular dynamics utilizing LSFM.
Abstract The AJCC/UICC clinical T staging for oral tongue squamous cell carcinoma (OTSCC) primarily relies on radiological depth of invasion (rDOI) and may not capture complex oral structure invasion (OSI). Therefore, we developed a magnetic resonance imaging (MRI)‐based risk stratification system integrating rDOI and OSI and explored the association between neoadjuvant chemoimmunotherapy (NACI) and overall survival (OS) across OSI classifications. This retrospective study included 1000 consecutive patients with nonmetastatic OTSCC who underwent curative‐intent surgery between 2010 and 2024. Pretreatment MRI scans were assessed for rDOI and invasion of 21 anatomical structures. Cox regression assessed prognostic associations; network and heatmap analyses mapped feature co‐occurrence. The primary endpoint was OS, and an exploratory analysis of NACI within OSI classification was performed in patients treated between 2020 and 2024. Four OSI categories were identified: very low, low, intermediate, and high risk. Compared with the very‐low‐risk group, mortality risk increased stepwise across the low‐, intermediate‐, and high‐risk groups, with adjusted hazard ratios of 3.62, 6.78, and 13.00, respectively. The OSI system provided modest but statistically significant incremental discrimination over clinical T staging (C‐index 0.669 vs. 0.633; p = .010), even after covariate adjustment (0.693 vs. 0.661; p = .016). In the 2020–2024 subgroup, NACI was associated with longer OS in the intermediate‐ and high‐risk OSI groups. The MRI‐based OSI system defines four anatomically interpretable risk groups, provides modest but statistically significant incremental discrimination over cT staging, and offers a framework for evaluating risk‐adapted strategies, including NACI.
ABSTRACT Gastrointestinal (GI) tumors represent a significant global health burden, necessitating robust strategies for early detection. However, the clinical utility of conventional protein biomarkers such as carcinoembryonic antigen (CEA), carbohydrate antigen 19‐9 (CA19‐9) in early diagnosis is restricted by insufficient sensitivity and specificity. In contrast, trefoil factor 3 (TFF3) is aberrantly expressed in GI tumors and holds potential as a GI tumor marker. This study enrolled 966 subjects and systematically evaluated the diagnostic performance of plasma TFF3, both independently and in combination with existing biomarkers, for gastric cancer (GC) and pan‐GI cancer. Results indicated significantly elevated plasma TFF3 levels in GC patients compared to healthy individuals ( p < .0001). In the GC cohort, a diagnostic panel combining TFF3, CEA, Pepsinogen I, Pepsinogen II, Gastrin‐17, age, and sex achieved an area under the curve (AUC) of 0.846. Furthermore, TFF3 demonstrated notable diagnostic efficacy across pan‐GI cancers. The AUC for differentiating between healthy control (HC) and colorectal cancer (CRC), esophageal cancer (EC), pancreatic cancer (PC), hepatocellular carcinoma (HCC), or pan‐GI cancer was 0.664, 0.774, 0.696, 0.714, and 0.741, respectively. Notably, combining TFF3 with CEA and baseline characteristics significantly improved the diagnostic performance in pan‐GI cancer, with an AUC of 0.825 and 0.816 in GC + EC and GC + EC + CRC respectively. These findings validated and clarified that plasma TFF3, in combination with different other markers such as gastric function indicators and CEA, holds promise as a biomarker for early screening and early diagnosis of GC and pan‐GI tumors, respectively.
Abstract Hepatocellular carcinoma (HCC) is a pernicious tumor with high heterogeneity and bleak prognosis, in which the immune microenvironment (IM) plays important roles. The underlying mechanisms of IM should be deeply investigated to discover novel therapeutic targets. In this study, we identified one long noncoding RNA associated with T cell activation, LINC00892, and deeply explored its association with the progression, prognosis, and the IM of HCC. We identified its interacted proteins and confirmed its regulatory mechanisms in HCC development. The dysregulated immune cell infiltration by LINC00892 was also identified. The results of bioinformatics analysis demonstrated that LINC00892 was downregulated in HCC tissues, and its high expression was connected with early TNM stage and better outcome. Further in vitro cellular experiments showed that LINC00892 overexpression significantly inhibited proliferation, migration, invasion, and colony formation abilities of HCC cells, which was also validated by in vivo animal model. Furthermore, high expression of LINC00892 was associated with incremental infiltration of CD8 + T cells, B cells, and macrophages, indicating its association with a better response to immunotherapy. Mechanistic analysis indicated that LINC00892 induces ribosomal stress by suppressing Shwachman‐Bodian Diamond Syndrome (SBDS) expression, further preventing the nuclear translocation of EIF6 and activating the p53 pathway, thereby inhibiting tumor progression. In summary, LINC00892 activates the p53 pathway through ribosomal stress mediated by SBDS/EIF6, regulates HCC progression, and probably enhances the infiltration of immune cells, showing potential application value as a prognostic indicator and immunotherapy orientation for HCC.
