Technol Health Care. 2025 May;33(3):1169-1181. doi: 10.1177/09287329241300986. Epub 2024 Dec 25.

ABSTRACT

A brain tumor (BT) is considered one of the most crucial and deadly diseases in the world, as it affects the central nervous system and its main functions. Headaches, nausea, and balance problems are caused by tumors pressing on nearby brain tissue and affecting its function. The existing techniques are challenging to analyze diseased brain images since abnormal brain tissues lead to distorted or biased results during image processing, like tissue segmentation and non-rigid registration. To overcome these issues, proposed a DS-GAN model for inpainting brain MRI images. Initially, the input MRI images are segmented using a Gated shape convolution neural network (GS-CNN). In the first GAN, grayscale pixel intensities and the remaining image edges are utilized to create edge generators or edge reconstruction Generative Adversarial Networks (EGAN), which are capable of creating false edges in areas that are missing. The results of the experimental results demonstrated that the Jaccard Index (JI) was 0.82, while the Dice Index (DI) was 0.86. The proposed DS-GAN in terms of L1 loss, PSNR, SSIM, and MSE obtained was 2.18, 0.972, 32.04, and 26.42. As compared to existing techniques, the proposed DS-GAN model achieves an overall accuracy of 99.18%.

PMID:40331553 | DOI:10.1177/09287329241300986

Technol Health Care. 2025 May;33(3):1199-1220. doi: 10.1177/09287329241302558. Epub 2024 Dec 29.

ABSTRACT

BackgroundEffective brain tumour therapy and better patient outcomes depend on early tumour diagnosis. Accurate diagnosis can be hampered by traditional imaging techniques' frequent struggles with low resolution and noise, especially in Low Dose CT scans.Objectives: Through the integration of deep learning methods and sophisticated image processing techniques, this study seeks to establish a novel framework, the Multi Layered Chroma Edge Deep Net (MLCED-Net), to improve the accuracy of brain tumour diagnosis in Low Dose CT images.MethodsUsing the Lucy-Richardson technique for picture deblurring, Adaptive Histogram Equalisation, and pixel normalization to lower noise and enhance image quality are some of the pre-processing stages that are part of the suggested strategy. Main characteristics from the processed pictures are then retrieved, including mean, energy, contrast, and entropy. Following the feeding of these characteristics, the MLCED-Net model is used for classification and segmentation tasks. It utilises a 15-layer deep learning architecture.ResultsThe MLCED-Net model outperformed previous techniques by achieving an amazing accuracy rate of 98.9% in the detection of brain tumours. The suggested procedures were effective, as seen by the significant increases in image quality that the Peak Signal-to-Noise Ratio (PSNR) values showed after post-processing.Conclusions: A reliable method for brain tumour diagnosis in low-dose CT scans is offered by the MLCED-Net framework's combination of multi-layered autoencoders, color-based operations, and edge detection techniques. The present work underscores the capacity of sophisticated deep learning models to augment diagnostic precision, hence augmenting patient care and results.

PMID:40331540 | DOI:10.1177/09287329241302558

NMR Biomed. 2025 Jun;38(6):e70057. doi: 10.1002/nbm.70057.

ABSTRACT

Quantitative susceptibility mapping (QSM) provides the spatial distribution of magnetic susceptibility within tissues through sequential steps: phase unwrapping and echo combination, mask generation, background field removal, and dipole inversion. Accurate mask generation is crucial, as masks excluding regions outside the brain and without holes are necessary to minimize errors and streaking artifacts during QSM reconstruction. Variations in susceptibility values can arise from different mask generation methods, highlighting the importance of optimizing mask creation. In this study, we propose QSMmask-net, a deep neural network-based method for generating precise QSM masks. QSMmask-net achieved the highest Dice score compared to other mask generation methods. Mean susceptibility values using QSMmask-net masks showed the lowest differences from manual masks (ground truth) in simulations and healthy controls (no significant difference, p > 0.05). Linear regression analysis confirmed a strong correlation with manual masks for hemorrhagic lesions (slope = 0.9814 ± 0.007, intercept = 0.0031 ± 0.001, R2 = 0.9992, p < 0.05). We have demonstrated that mask generation methods can affect the susceptibility value estimations. QSMmask-net reduces the labor required for mask generation while providing mask quality comparable to manual methods. The proposed method enables users without specialized expertise to create optimized masks, potentially broadening QSM applicability efficiently.

PMID:40331503 | DOI:10.1002/nbm.70057

Int J Mol Sci. 2025 Apr 18;26(8):3850. doi: 10.3390/ijms26083850.

ABSTRACT

The resolution of the recent COVID-19 pandemic still requires attention, since the consequences of having suffered the infection, even in mild cases, are associated with several acute and chronic pathological conditions referred to as post-COVID syndrome (PCS). PCS often manifests with pulmonary disease and, in up to 9% of cases, a more serious complication known as post-COVID-19 pulmonary fibrosis (PC19-PF), which has a similar clinical course as idiopathic pulmonary fibrosis (IPF). Generating knowledge to provide robust evidence about the clinical benefits of different therapeutic strategies to treat the pulmonary effects of PCS can provide new insights to amplify therapeutic options for these patients. We present evidence found after a scoping review, following extended PRIMSA guidelines, for the use of immunomodulators in pulmonary PCS. We start with a brief description of the immunomodulatory properties of the relevant drugs, their clinically proven efficacy for viral infections and chronic inflammatory conditions, and their use during the COVID-19 pandemic. We emphasize the need for well-designed clinical trials to improve our understanding the physiopathology of pulmonary PCS and PC19-PF and also to determine the efficacy and safety of candidate treatments.

PMID:40332501 | DOI:10.3390/ijms26083850

Insects. 2025 Apr 17;16(4):422. doi: 10.3390/insects16040422.

ABSTRACT

The Neotropical brown stink bug (NBSB), Euschistus heros, is the most prevalent sucking soybean pest in Brazil, and control of it largely relies on the application of synthetic insecticides such as ethiprole, a phenylpyrazole insecticide targeting GABA-gated chloride channels encoded by the Rdl (resistant to dieldrin) gene. This study monitored 41 NBSB populations collected between 2021 and 2024 and revealed, for the first time, the presence of a mutation, A301S, in NBSB RDL receptors commonly known to confer target-site resistance to channel blockers such as phenylpyrazoles. Laboratory contact bioassays with ethiprole at 150 g a.i./ha (ethiprole label dose) revealed that most populations were quite susceptible, despite rather high resistance allele frequencies in some populations. Genotyping results confirmed that susceptible and A301S heterozygous genotypes largely dominate in frequency compared to homozygous resistant individuals, which exhibited high survivorship (84%) when exposed to discriminating rates of ethiprole in laboratory bioassays, while susceptible and heterozygote individuals showed lower survival rates (13% and 34%, respectively), suggesting an incompletely recessive trait conferring ethiprole resistance. Furthermore, we developed a TaqMan assay for molecular genotyping to monitor the spread of resistance allele frequency and to inform resistance management strategies for sustainable NBSB control using highly effective phenylpyrazole insecticides such as ethiprole.

PMID:40332961 | DOI:10.3390/insects16040422

Int J Mol Sci. 2025 Apr 15;26(8):3739. doi: 10.3390/ijms26083739.

ABSTRACT

As human life expectancy continues to rise, managing age-related diseases and preserving health in later years remain significant challenges. Consequently, there is a growing demand for strategies that enhance both the quality and the duration of life. Interventions that promote longevity, particularly those derived from natural sources, are popular for their potential to address age-related health concerns. Adaptogens-herbs, roots, and mushrooms-are valued in food science and nutrition for their ability to enhance resilience and overall well-being. Among these, Rhaponticum carthamoides (Willd.) Iljin, known as maral root (Russian leuzea), holds a prominent place in Siberian traditional medicine. The root extract, abundant in bioactive compounds such as flavonoids and phytoecdysteroids, is reputed for reducing fatigue, boosting strength, and offering immunomodulatory benefits. However, the effects of the plant extract on lifespan and age-related decline remains poorly studied. This study investigates the effect of maral root extract and phytoecdysteroids-ecdysterone, ponasterone, and turkesterone-on aging using Caenorhabditis elegans as a model organism. A sensitive liquid chromatography method with photodiode array detection was developed and validated to quantify the phytoecdysteroids in the extract. Behavioural and stress-response assays revealed that maral root not only extends lifespan but also significantly enhanced healthspan, stress resilience, and fitness in the nematodes. Additionally, treatment with ecdysterone, the most abundant compound in the root extract, improved healthspan by enhancing stress response. These findings underscore the potential of maral root as a natural adaptogen to mitigate age-related decline, providing valuable insights into natural longevity interventions.

PMID:40332350 | DOI:10.3390/ijms26083739

Acc Chem Res. 2025 May 7. doi: 10.1021/acs.accounts.5c00117. Online ahead of print.

ABSTRACT

ConspectusMethodological development in the fields of genetics, chemical biology, and biochemistry over the last several decades has provided researchers with a diverse set of powerful tools to investigate biological processes. Leveraging these innovations in concert, scientists can now characterize biological pathways at a level of complexity ranging from systems biology down to molecular and atomic detail.Throughout this Account, we illustrate how discoveries made using these tools build on each other to develop a comprehensive understanding of biological pathways. Advancements in genetic sequencing facilitates association of genotypes and phenotypes, independent of biochemical mechanism. Through the biochemical reconstitution of the interactions between biological macromolecules─including the small molecules (ligands and metabolites) and proteins─that participate in these biological pathways, scientists can characterize the specific molecular features that link genotype and phenotype. This facilitates identification of targets within these pathways that can be manipulated to achieve a greater understanding of the biological process or to develop interventions to improve human health outcomes.Specifically, we describe how this toolbox was leveraged to discover and characterize the molecular biochemistry underlying control of pathogenicity in the Gram-positive bacterium Staphylococcus aureus. Concurrent with advancements in the investigative tools available to the scientific community, we and others reported on the genetic, molecular, and biochemical/biophysical components of this regulatory system. Virulence control in S. aureus is achieved through a chemical system of bacterial cell-to-cell communication indexed to local population density, referred to as quorum sensing (QS). We and our collaborators identified that this QS system is encoded in the accessory gene regulator (agr) operon and functions via the biosynthesis, secretion, and accumulation of a short peptide signaling molecule─the autoinducing peptide (AIP)─in the local environment correlated with the growth of S. aureus in the same biological niche. Above a threshold concentration, these AIPs bind and activate a cell-surface receptor to stimulate an intracellular response resulting in altered gene expression and bacterial group behaviors. We discovered that chemical modification of these AIPs often generates molecules that exhibit potent inhibition of agr QS, with demonstrated therapeutic potential to treat S. aureus infections. We went on to characterize the biochemical mechanism of signaling molecule biosynthesis and receptor activation in controlled systems through in vitro reconstitution of the constituent enzymes and substrates. Biochemical reconstitution enabled quantitative assessment of biophysical parameters. These efforts culminated in the comprehensive characterization and functional in vitro reconstitution of agr QS in a synthetic system in a minimal model at the interface of genotype, mechanism, and phenotype.

PMID:40331756 | DOI:10.1021/acs.accounts.5c00117

Nucleic Acids Res. 2025 May 7:gkaf369. doi: 10.1093/nar/gkaf369. Online ahead of print.

ABSTRACT

Finding appropriate software and parameter settings to process shotgun metagenome data is essential for meaningful metagenomic analyses. To enable objective and comprehensive benchmarking of metagenomic software, the community-led initiative for the Critical Assessment of Metagenome Interpretation (CAMI) promotes standards and best practices. Since 2015, CAMI has provided comprehensive datasets, benchmarking guidelines, and challenges. However, benchmarking had to be conducted offline, requiring substantial time and technical expertise and leading to gaps in results between challenges. We introduce the CAMI Benchmarking Portal-a central repository of CAMI resources and web server for the evaluation and ranking of metagenome assembly, binning, and taxonomic profiling software. The portal simplifies evaluation, enabling users to easily compare their results with previous and other users' submissions through a variety of metrics and visualizations. As a demonstration, we benchmark software performance on the marine dataset of the CAMI II challenge. The portal currently hosts 28 675 results and is freely available at https://cami-challenge.org/.

PMID:40331433 | DOI:10.1093/nar/gkaf369

Front Aging Neurosci. 2025 Apr 22;17:1549072. doi: 10.3389/fnagi.2025.1549072. eCollection 2025.

ABSTRACT

OBJECTIVES: To investigate whether cerebrospinal fluid (CSF) circulation markers alter in patients with probable cerebral amyloid angiopathy (pCAA) and whether they are associated with brain degeneration and cognitive impairment.

METHODS: We screened pCAA patients from the ADNI3 database according to the Boston 2.0 Criteria. Fifty-two patients with cognitive impairment (26 pCAA; 26 age-sex-matched non-pCAA) and 26 age-sex-matched cognitively normal control (NC) were included in this study. All participants underwent neurological MRI and cognitive assessments. Choroid plexus (ChP) was segmented using a deep learning-based method and its volume was extracted. Diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) was used to assess perivenous fluid mobility. AD pathological markers (Aβ and tau) were assessed using positron emission tomography. Brain parenchymal damage markers included white matter hyperintensities (WMH) volume and brain atrophy ratio. All markers were compared among the three groups. Correlations among the ChP volume, DTI-ALPS index, parenchymal damage markers, and cognitive scales were analyzed in the pCAA group.

RESULTS: The three groups exhibited significant differences in cognitive scores, AD biomarkers, and imaging markers. Post hoc analyses showed that patients with pCAA had significantly higher WMH volume, higher Aβ and tau deposition, and lower DTI-ALPS compared to NC. However, no difference in ChPs volume was found among the groups. Controlling for age, sex, and vascular risk factors, partial correlation analyses showed a significant negative correlation between the DTI-ALPS and WMH volume fraction (r = -0.606, p = 0.002). ChP volume was significantly associated with the Montreal cognitive assessment score (r = -0.492, p = 0.028).

CONCLUSION: CSF circulation markers were associated with elevated WMH burden and cognitive impairments in probable CAA.

PMID:40330595 | PMC:PMC12053238 | DOI:10.3389/fnagi.2025.1549072

Circ Res. 2025 May 7. doi: 10.1161/CIRCRESAHA.124.325699. Online ahead of print.

ABSTRACT

BACKGROUND: It has become evident that some women develop preeclampsia despite aspirin. This study aimed to examine how such aspirin nonresponsiveness develops in high-risk preeclampsia pregnancies by exploring the role of genetic polymorphisms and aspirin metabolism.

METHODS: The study involved pregnant women who developed preeclampsia despite low-dose aspirin and those who did not. First, we conducted a pharmacogenomic association study exploring the association of potential genetic variants with aspirin nonresponsiveness. Next, we analyzed the rate of enzymatic aspirin hydrolysis in maternal plasma. The extent of placental exposure to acetylsalicylic acid and its bioactive metabolites, that is, salicylic acid and gentisic acid, was determined by liquid chromatography-mass spectrometry. The expressions of AMEs (aspirin metabolizing enzymes), that is, GLYAT (glycine-N-acyltransferase), UGT1A6, CYP2E1, and NAT2 in the placenta, were analyzed by quantitative reverse transcription polymerase chain reaction, immunohistochemistry staining, and ELISA. Finally, the effects of AMEs were further examined on HTR-8/SVneo and human primary cytotrophoblast cells.

RESULT: Our genetic study showed that single-nucleotide polymorphisms (SNPs) of genes involved in aspirin pharmacokinetics and pharmacodynamics were not associated with aspirin nonresponsiveness in preeclampsia. Rates of aspirin hydrolysis in maternal plasma and the concentrations of acetylsalicylic acid, salicylic acid, and gentisic acid in the placenta did not differ between aspirin-responsive and aspirin-nonresponsive women. Intriguingly, GLYAT was significantly upregulated in the aspirin-nonresponsive placenta and associated with aspirin nonresponsiveness. This overexpression of GLYAT was found to diminish the proangiogenic, anti-inflammatory, and antisenescence effects of salicylic acid in HTR-8/SVneo and human primary cytotrophoblast cells.

CONCLUSIONS: Our study revealed that maternal genetic factors and plasma aspirin hydrolysis are not among the decisive factors in determining the effectiveness of low-dose aspirin in preventing preeclampsia among high-risk women. Instead, placental GLYAT appears to play a key role by limiting the effect of salicylic acid in the placenta.

PMID:40329906 | DOI:10.1161/CIRCRESAHA.124.325699

Acta Biochim Biophys Sin (Shanghai). 2025 May 6. doi: 10.3724/abbs.2025052. Online ahead of print.

ABSTRACT

Platinum drugs are widely used in lung cancer chemotherapy, but the immune characteristics of different individuals have different effects on the sensitivity and side effects of platinum drugs. In this study, we use 731 kinds of immune cell traits of 3757 healthy individuals and 429 patients with non-small cell lung cancer (NSCLC) in Xiangya Hospital of Central South University to conduct a Mendel randomized analysis in order to find out the causal relationship between some immune cell traits and the efficacy and adverse reactions of platinum drugs. We find that CD19 on CD24 +CD27 + B cell (OR = 0.598, P = 0.004) is the most significant immune cell trait as the protective factor of efficacy. HLA-DR +CD8 + T cell % lymphocyte (OR = 0.427, P = 7.55 × 10 -4) and HLA-DR +CD8 + T cell % T cell (OR = 0.471, P = 0.003) are the protective factors of liver injury. CD39 on CD39 + secreting CD4 + regulatory T cell (OR = 28.729, P = 0.009) and CD3 on CD39 + resting CD4 regulatory T cell (OR = 3.024, P = 0.009) are the risk factors of renal injury. Meanwhile, B cell-related traits mainly affect gastrointestinal upset and cutaneous toxicity, while T cell-related traits mainly affect other outcome variables. These findings may promote our understanding of the relationship between the efficacy and adverse reactions of platinum drugs and the immune system, and promote future development of biomarkers for predicting the efficacy and adverse reactions of platinum drugs.

PMID:40329806 | DOI:10.3724/abbs.2025052

Front Immunol. 2025 Apr 22;16:1576141. doi: 10.3389/fimmu.2025.1576141. eCollection 2025.

ABSTRACT

In recent years, the study of the interaction between gut microbiota and distant organs such as the heart, lungs, brain, and liver has become a hot topic in the field of gut microbiology. With a deeper understanding of its immune regulation and mechanisms of action, these findings have increasingly highlighted their guiding value in clinical practice. The gut is not only the largest digestive organ in the human body but also the habitat for most microorganisms. Imbalances in gut microbial communities have been associated with various lung diseases, such as allergic asthma and cystic fibrosis. Furthermore, gut microbial communities have significant impacts on metabolic function and immune responses. Their metabolites not only regulate gastrointestinal immune systems but may also affect distant organs such as the lungs and brain. As one of the most common types of respiratory system diseases worldwide, pulmonary infections have high morbidity and mortality rates. Pulmonary infections caused by immune dysfunction can lead to gastrointestinal problems like diarrhea, further resulting in imbalances within complex interactions that are associated with abnormal manifestations under disequilibrium conditions. Meanwhile, clinical interventions can significantly modulate the composition of gut microbiota, and alteration in gut microbiota may subsequently indicate susceptibility to pulmonary infections and even contribute to the prevention or regulation of their progression. This review delves into the interaction between gut microbiota and pulmonary infections, elucidating the latest advancements in gut-lung axis research and providing a fresh perspective for the treatment and prevention of pneumonia.

PMID:40330490 | PMC:PMC12052896 | DOI:10.3389/fimmu.2025.1576141

Int J Cardiol Congenit Heart Dis. 2025 Apr 15;20:100584. doi: 10.1016/j.ijcchd.2025.100584. eCollection 2025 Jun.

ABSTRACT

BACKGROUND: High-throughput proteomics platforms using aptamers (SomaScan) or proximity extension assay (Olink) provide novel opportunities for improving diagnostic and risk stratification tools in cardiovascular diseases, including understudied congenital heart diseases. The correlation between these proteomics approaches has not yet been studied among individuals with a Fontan circulation.

OBJECTIVE: The correlation of plasma protein measurements between SomaScan and Olink platforms was evaluated in adults with a Fontan circulation.

METHODS: We measured 491 proteins in plasma of 71 adults with a Fontan circulation using Olink and SomaScan. Missing Olink measurements (0.13%, 47/34,861) were imputed using non-parametric imputation. Spearman's rank correlation coefficient for absolute values of protein expression between platforms was calculated. Protein correlation frequencies were compared to 3 cohorts reported in the literature using Pearson's Chi-squared test of independence.

RESULTS: Overall, protein correlations between Olink and SomaScan measurements were moderately strong for most proteins, (rho > 0.4 for 57.2%), but with substantial variability (median correlation = 0.457, IQR = 0.538). The distribution of protein correlations was qualitatively similar to published literature in non-Fontan cohorts. Both Olink and SomaScan identified proteins with sex-based differences; both identified differences in myostatin and leptin, but each identified additional nonoverlapping sexually dimorphic proteins (n = 14 Olink, n = 5 SomaScan).

CONCLUSIONS: In adults with a Fontan circulation, correlations between plasma proteins measured by Olink and SomaScan varied widely, approximately in line with prior reports in other populations. While these tools may be uniquely useful to generate hypotheses, specifically regarding potential molecular mechanisms, more definitive inference requires independent validation.

PMID:40330320 | PMC:PMC12053979 | DOI:10.1016/j.ijcchd.2025.100584

Cyberpsychol Behav Soc Netw. 2025 May 7. doi: 10.1089/cyber.2024.0457. Online ahead of print.

ABSTRACT

Emotion detection is crucial in various domains, including psychology, health, social sciences, and marketing. Specifically, in psychology, identifying negative emotions in short Spanish texts, such as tweets, is vital for understanding individuals' emotional states. However, this process is challenging because of factors such as lack of context, cultural nuances, and ambiguous expressions. Although much research on emotion classification in tweets has focused on applications such as crisis analysis, mental health monitoring, and affective computing, most of it has been conducted in English, leaving a significant gap in addressing the emotional needs of Spanish-speaking communities. To address this gap, we used a corpus of 12,000 Spanish tweets tagged with Ekman's negative emotions (sadness, anger, fear, and disgust). Traditional features (n-grams of different types and sizes), syntactic n-grams, and combined features were evaluated. Different deep neural networks, including convolutional neural networks, Bidirectional Encoder Representations of Transformers (BERT), and the robust optimized BERT approach called RoBERTa, were implemented and compared with traditional machine learning methods to identify the most effective method. Extensive testing revealed that BERT achieved the best result, with a macro F1 score of 0.9973. Furthermore, we reported the carbon emissions generated during the training of each implemented method. This study makes a unique contribution by focusing on negative emotions in Spanish, leveraging one of the largest and highest-quality corpora available. It stands out for implementing advanced transformers such as RoBERTa and integrating combined and syntactic n-grams in traditional methods. Furthermore, it highlights how parameters, features, and preprocessing significantly influence performance.

PMID:40331318 | DOI:10.1089/cyber.2024.0457

J Hip Preserv Surg. 2024 Dec 12;12(1):27-32. doi: 10.1093/jhps/hnae041. eCollection 2025 Jan.

ABSTRACT

Deep learning is revolutionizing medical imaging analysis by enabling the classification of various pathoanatomical conditions at scale. Unfortunately, there have been a limited number of accurate and efficient machine learning (ML) algorithms that have been developed for the diagnostic workup of morphological hip pathologies, including developmental dysplasia of the hip and femoroacetabular impingement. The current study reports on the performance of a novel ML model with YOLOv5 and ConvNeXt-Tiny architecture in predicting the morphological features of these conditions, including cam deformity, ischial spine sign, dysplastic appearance, and other abnormalities. The model achieved 78.0% accuracy for detecting cam deformity, 87.2% for ischial spine sign, 76.6% for dysplasia, and 71.6% for all abnormalities combined. The model achieved an Area under the Receiver Operating Curve of 0.89 for ischial spine sign, 0.80 for cam deformity, 0.80 for dysplasia, and 0.81 for all abnormalities combined. Inter-rater agreement among surgeons, assessed using Gwet's AC1, was substantial for dysplasia (0.83) and all abnormalities (0.88), and moderate for ischial spine sign (0.75) and cam deformity (0.61).

PMID:40331073 | PMC:PMC12051864 | DOI:10.1093/jhps/hnae041

Photoacoustics. 2025 Apr 18;43:100723. doi: 10.1016/j.pacs.2025.100723. eCollection 2025 Jun.

ABSTRACT

Three-dimensional (3D) photoacoustic imaging (PAI) with detector arrays has shown superior imaging capabilities in biomedical applications. However, the quality of 3D PAI is often degraded due to reconstruction artifacts caused by sparse detectors. Existing iterative or deep learning-based methods are either time-consuming or require large training datasets, limiting their practical application. Here, we propose Zero-Shot Artifact2Artifact (ZS-A2A), a zero-shot self-supervised artifact removal method based on a super-lightweight network, which leverages the fact that patterns of artifacts are more sensitive to sensor data loss. By randomly dropping acquired PA data, it spontaneously generates subset data to reconstruct images, which in turn stimulates the network to learn the artifact patterns in reconstruction results, thus enabling zero-shot artifact removal. This approach requires neither training data nor prior knowledge of the artifacts, making it suitable for artifact removal for arbitrary detector array configurations. We validated ZS-A2A in both simulation study and i n v i v o animal experiments. Results demonstrate that ZS-A2A achieves high performance compared to existing zero-shot methods.

PMID:40331014 | PMC:PMC12051505 | DOI:10.1016/j.pacs.2025.100723

iScience. 2025 Apr 9;28(5):112363. doi: 10.1016/j.isci.2025.112363. eCollection 2025 May 16.

ABSTRACT

Despite wide variation, each cell type has an optimal size. Maintaining optimal size is essential for cellular fitness and function but the biological basis for this remains elusive. Here, we performed fitness analysis involving genome-wide CRISPR-Cas9 knockout data from tens of human cell lines and identified that cell size influences the essentiality of genes related to mitochondria and membrane repair. These genes also included glutathione peroxidase 4 (GPX4), which safeguards membranes from oxidative damage and prevents ferroptosis-iron-dependent death. Growth beyond normal size, with or without cell-cycle arrest, increased lipid peroxidation, resulting in a ferroptosis-sensitive state. Proteomic analysis revealed cell-cycle-independent superscaling of endoplasmic reticulum, accumulation of iron, and lipidome remodeling. Even slight increases from normal cell size sensitized proliferating cells to ferroptosis as evidenced by deep-learning-based single-cell analysis. Thus, lipid peroxidation may be a fitness trade-off that constrains cell enlargement and contributes to the establishment of an optimal cell size.

PMID:40330887 | PMC:PMC12053632 | DOI:10.1016/j.isci.2025.112363

Proc IEEE Int Conf Acoust Speech Signal Process. 2023 Jun;2023. doi: 10.1109/icassp49357.2023.10096985. Epub 2023 May 5.

ABSTRACT

Deep learning based speech enhancement has achieved remarkable success, but challenges remain in low signal-to-noise ratio (SNR) nonstationary noise scenarios. In this study, we propose to incorporate diffusion-based learning into an enhancement model and improve robustness in extremely noisy conditions. Specifically, a frequency-domain diffusion-based generative module is employed, and it accepts the enhanced signal obtained from a time-domain supervised enhancement module as an auxiliary input to learn to recover clean speech spectrograms. Experimental results on the TIMIT dataset demonstrate the advantage of this approach and show better enhancement performance over other strong baselines in both -5 and -10 dB SNR noisy conditions.

PMID:40330790 | PMC:PMC12051501 | DOI:10.1109/icassp49357.2023.10096985

Health Sci Rep. 2025 May 5;8(5):e70802. doi: 10.1002/hsr2.70802. eCollection 2025 May.

ABSTRACT

PURPOSE: Alzheimer's disease (AD) is a severe neurological disease that significantly impairs brain function. Timely identification of AD is essential for appropriate treatment and care. This comprehensive review intends to examine current developments in deep learning (DL) approaches with neuroimaging for AD diagnosis, where popular imaging types, reviews well-known online accessible data sets, and describes different algorithms used in DL for the correct initial evaluation of AD are presented.

SIGNIFICANCE: Conventional diagnostic techniques, including medical evaluations and cognitive assessments, usually not identify the initial stages of Alzheimer's. Neuroimaging methods, when integrated with DL techniques, have demonstrated considerable potential in enhancing the diagnosis and categorization of AD. DL models have received significant interest due to their capability to identify AD in its early phases automatically, which reduces the mortality rate and treatment cost of AD.

METHOD: An extensive literature search was performed in leading scientific databases, concentrating on papers published from 2021 to 2025. Research leveraging DL models on different neuroimaging techniques such as magnetic resonance imaging (MRI), positron emission tomography, and functional magnetic resonance imaging (fMRI), and so forth. The review complies with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

RESULTS: Current developments show that CNN-based techniques, especially those utilizing hybrid and transfer learning frameworks, outperform conventional DL methods. Research employing the combination of multimodal neuroimaging data has demonstrated enhanced diagnostic precision. Still, challenges such as method interpretability, data heterogeneity, and limited data exist as significant issues.

CONCLUSION: DL has considerably improved the accuracy and reliability of AD diagnosis with neuroimaging. Regardless of issues with data accessibility and adaptability, current studies into the interpretability of models and multimodal fusion provide potential for clinical application. Further research should concentrate on standardized data sets, rigorous validation architectures, and understandable AI methodologies to enhance the effectiveness of DL methods in AD prediction.

PMID:40330773 | PMC:PMC12051440 | DOI:10.1002/hsr2.70802

Health Sci Rep. 2025 May 5;8(5):e70799. doi: 10.1002/hsr2.70799. eCollection 2025 May.

ABSTRACT

BACKGROUND AND AIMS: Stroke remains a leading cause of mortality and long-term disability worldwide, presenting a significant global health challenge. Effective early prediction models are essential for reducing its impact. This study introduces a novel ensemble method for predicting stroke using two datasets: a primary dataset collected from a hospital, containing medical histories and clinical parameters, and a secondary dataset.

METHODS: We applied several preprocessing techniques, including outlier detection, data normalization, k-means clustering, and missing value detection, to refine the datasets. A novel ensemble classifier was developed, combining AdaBoost, Gradient Boosting Machine (GBM), Multilayer Perceptron (MLP), and Random Forest (RF) algorithms to enhance predictive accuracy. Additionally, Explainable Artificial Intelligence (XAI) techniques such as SHAP and LIME were integrated to elucidate key features influencing stroke prediction.

RESULTS: The proposed ensemble classifier achieved an accuracy of 95% for the secondary dataset and 80.36% for the primary dataset. Comparative analysis with other machine learning models highlighted the superior performance of the ensemble approach. The integration of XAI further provided insights into the critical indicators influencing stroke classification, improving model interpretability and decision-making.

CONCLUSION: Our study demonstrates that the novel ensemble classifier, supported by effective preprocessing and XAI techniques, is a powerful tool for stroke prediction. The high accuracy rates achieved validate its effectiveness and potential for practical clinical application. Future work will focus on incorporating deep learning techniques and medical imaging to further improve classification accuracy and model performance.

PMID:40330769 | PMC:PMC12052519 | DOI:10.1002/hsr2.70799