Comput Biol Med. 2025 Mar 28;190:110019. doi: 10.1016/j.compbiomed.2025.110019. Online ahead of print.

ABSTRACT

Communication involves exchanging information between individuals or groups through various media sources. However, limitations such as hearing loss can make it difficult for some individuals to understand the information delivered during speech communication. Conventional methods, including sign language, written text, and manual lipreading, offer some solutions; however, emerging software-based tools using artificial intelligence (AI) are introducing more effective approaches. Many approaches rely on AI to improve communication quality, with the current trend of leveraging deep learning being a particularly effective tool. This paper presents a comprehensive Systematic Literature Review (SLR) of research trends in automatic lipreading technologies, a critical field in enhancing communication among individuals with hearing impairments. The SLR, which followed the Preferred Reporting Items for Systematic Literature Review and Meta-Analysis (PRISMA) protocol, identified 114 original research articles published between 2014 and mid-2024. The essential information from these articles was summarized, including the trends in automatic lipreading research, dataset availability, task categories, existing approaches, and architectures for automatic lipreading systems. The results showed that various techniques and advanced deep learning models achieved convincing performance to become state-of-the-art in automatic lipreading tasks. However, several challenges, such as insufficient data quantity, inadequate environmental conditions, and language diversity, must be resolved in the future. Furthermore, many improvements have been made to the deep learning models to overcome these challenges and become a massive solution, particularly for automatic lipreading tasks in the near future.

PMID:40157316 | DOI:10.1016/j.compbiomed.2025.110019

Comput Biol Med. 2025 Mar 28;190:110048. doi: 10.1016/j.compbiomed.2025.110048. Online ahead of print.

ABSTRACT

BACKGROUND AND OBJECTIVE: Accurate medical tumor segmentation is critical for early diagnosis and treatment planning, significantly improving patient outcomes. This study aims to enhance liver and tumor segmentation from CT and liver images by developing a novel model, ResTransUNet, which combines convolutional and transformer blocks to improve segmentation accuracy.

METHODS: The proposed ResTransUNet model is a custom implementation inspired by the TransUNet architecture, featuring a Standalone Transformer Block and ResNet50 as the backbone for the encoder. The hybrid architecture leverages the strengths of Convolutional Neural Networks (CNNs) and Transformer blocks to capture both local features and global context effectively. The encoder utilizes a pre-trained ResNet50 to extract rich hierarchical features, with key feature maps to preserved it as skip connections. The Standalone Transformer Block, integrated into the model, employs multi-head attention mechanisms to capture long-range dependencies across the image, enhancing segmentation performance in complex cases. The decoder reconstructs the segmentation mask by progressively upsampling encoded features while integrating skip connections, ensuring both semantic information and spatial details are retained. This process culminates in a precise binary segmentation mask that effectively distinguishes liver and tumor regions.

RESULTS: The ResTransUNet model achieved superior Dice Similarity Coefficient (DSC) for liver segmentation (98.3% on LiTS and 98.4% on 3D-IRCADb-01) and for tumor segmentation from CT images (94.7% on LiTS and 89.8% on 3D-IRCADb-01) as well as from liver images (94.6% on LiTS and 91.1% on 3D-IRCADb-01). The model also demonstrated high precision, sensitivity, and specificity, outperforming current state-of-the-art methods in these tasks.

CONCLUSIONS: The ResTransUNet model demonstrates robust and accurate performance in complex medical image segmentation tasks, particularly in liver and tumor segmentation. These findings suggest that ResTransUNet has significant potential for improving the precision of surgical interventions and therapy planning in clinical settings.

PMID:40157314 | DOI:10.1016/j.compbiomed.2025.110048

Commun Eng. 2025 Mar 29;4(1):59. doi: 10.1038/s44172-025-00399-1.

NO ABSTRACT

PMID:40158063 | DOI:10.1038/s44172-025-00399-1

Sci Rep. 2025 Mar 29;15(1):10897. doi: 10.1038/s41598-025-95633-3.

ABSTRACT

Lung cancer is the leading cause of cancer-related deaths globally, with radiotherapy as a key treatment modality for inoperable cases. Lactate, once considered a by-product of anaerobic cellular metabolism, is now considered critical for cancer progression. Lactate dehydrogenase B (LDHB) converts lactate to pyruvate and supports mitochondrial metabolism. In this study, a re-analysis of our previous transcriptomic data revealed that LDHB silencing in the NSCLC cell lines A549 and H358 dysregulated 1789 genes, including gene sets associated with cell cycle and DNA repair pathways. LDHB silencing increased H2AX phosphorylation, a surrogate marker of DNA damage, and induced cell cycle arrest at the G1/S or G2/M checkpoint depending on the p53 status. Long-term LDHB silencing sensitized A549 cells to radiotherapy, resulting in increased DNA damage and genomic instability as evidenced by increased H2AX phosphorylation levels and micronuclei accumulation, respectively. The combination of LDHB silencing and radiotherapy increased protein levels of the senescence marker p21, accompanied by increased phosphorylation of Chk2, suggesting persistent DNA damage. Metabolomics analysis revealed that LDHB silencing decreased nucleotide metabolism, particularly purine and pyrimidine biosynthesis, in tumor xenografts. Nucleotide supplementation partially attenuated DNA damage caused by combined LDHB silencing and radiotherapy. These findings suggest that LDHB supports metabolic homeostasis and DNA damage repair in NSCLC, while its silencing enhances the effects of radiotherapy by impairing nucleotide metabolism and promoting persistent DNA damage.

PMID:40158058 | DOI:10.1038/s41598-025-95633-3

Sci Rep. 2025 Mar 29;15(1):10856. doi: 10.1038/s41598-025-95738-9.

ABSTRACT

SF3B1 gene mutations are prevalent in myelodysplastic syndrome (MDS) and define a distinct disease subtype. These mutations are associated with dysregulated genes and pathways, offering potential for novel therapeutic approaches. However, the aberrant mRNA alternative splicing landscape in SF3B1-deficient MDS cells remains underexplored. In this study, we investigated the influence of SF3B1 gene alterations on the pre-mRNA splicing landscape in MDS cells using transcriptomic data from two independent MDS cohorts. we identified over 5000 significant differential alternative splicing events associated with SF3B1 mutation. This work corroborates previous studies, showing significant enrichment of MYC activity and heme metabolism in SF3B1 mutant cells. A key novel finding of this study is the identification of a gene expression signature driven by SF3B1 mutations, centered on protein post-translational modifications. Notably, we discovered aberrant alternative splicing of the tumor suppressor gene UBA7, leading to significantly reduced gene expression. This dysregulation implicates UBA7 as a critical player in MDS pathogenesis. Importantly, the clinical relevance of this finding is underscored by the observation that low UBA7 gene expression was associated with poor overall survival in chronic lymphocytic leukemia (CLL), another hematological malignancy with frequent SF3B1 mutations. Furthermore, a similar association between low UBA7 gene expression and poor survival outcomes was observed across multiple tumor types in the TCGA database, highlighting the broader implications of UBA7 dysregulation in cancer biology. These findings provide new insights into the mechanisms by which SF3B1 mutations reshape the pre-mRNA splicing landscape and drive disease pathogenesis in MDS. Furthermore, they underscore the potential of UBA7 as a biomarker to stratify SF3B1-mutant MDS and CLL patients, offering a refined approach for risk assessment and highlighting opportunities for targeted therapeutic interventions.

PMID:40158006 | DOI:10.1038/s41598-025-95738-9

Nat Commun. 2025 Mar 29;16(1):3065. doi: 10.1038/s41467-025-58259-7.

ABSTRACT

Enzymes are essential catalysts in biological systems. Substrate inhibition, once dismissed, is now observed in 20% of enzymes1 and is attributed to the formation of an unproductive enzyme-substrate complex, with no structural evidence of unproductivity provided to date1-6. This study uncovers the molecular mechanism of substrate inhibition in tobacco glucosyltransferase NbUGT72AY1, which transfers glucose to phenols for plant protection. The peculiarity that β-carotene strongly attenuates the substrate inhibition of NbUGT72AY1, despite being a competitive inhibitor, allows to determine the conformational changes that occur during substrate binding in both active and substrate-inhibited complexes. Crystallography reveals structurally different ternary enzyme-substrate complexes that do not conform to classical mechanisms. An alternative pathway suggests substrates bind randomly, but the reaction occurs only if a specific order is followed (asymmetric cooperativity). This unreported paradigm explains substrate inhibition and reactivation by competitive inhibitors, opening new research avenues in metabolic regulation and industrial applications.

PMID:40157902 | DOI:10.1038/s41467-025-58259-7

Biotechnol Adv. 2025 Mar 27:108568. doi: 10.1016/j.biotechadv.2025.108568. Online ahead of print.

ABSTRACT

Curcumin, a natural compound found in turmeric, has shown promise in treating brain-related diseases and conditions associated with aging. Curcumin has shown multiple anti-inflammatory and brain-protective effects, but its clinical use is limited by challenges like poor absorption, specificity and delivery to the right tissues. A range of contemporary approaches at the intersection with bioengineering and systems biology are being explored to address these challenges. Data from preclinical and human studies highlight various neuroprotective actions of curcumin, including the inhibition of neuroinflammation, modulation of critical cellular signaling pathways, promotion of neurogenesis, and regulation of dopamine levels. However, curcumin's multifaceted effects - such as its impact on microRNAs and senescence markers - suggest novel therapeutic targets in neurodegeneration. Tetrahydrocurcumin, a primary metabolite of curcumin, also shows potential due to its presence in circulation and its anti-inflammatory properties, although further research is needed to elucidate its neuroprotective mechanisms. Recent advancements in delivery systems, particularly brain-targeting nanocarriers like polymersomes, micelles, and liposomes, have shown promise in enhancing curcumin's bioavailability and therapeutic efficacy in animal models. Furthermore, the exploration of drug-laden scaffolds and dermal delivery may extend the pharmacological applications of curcumin. Studies reviewed here indicate that engineered dermal formulations and devices could serve as viable alternatives for neuroprotective treatments and to manage skin or musculoskeletal inflammation. This work highlights the need for carefully designed, long-term studies to better understand how curcumin and its bioactive metabolites work, their safety, and their effectiveness.

PMID:40157560 | DOI:10.1016/j.biotechadv.2025.108568

Gastroenterology. 2025 Mar 27:S0016-5085(25)00540-2. doi: 10.1053/j.gastro.2025.03.018. Online ahead of print.

NO ABSTRACT

PMID:40157433 | DOI:10.1053/j.gastro.2025.03.018

Cell Metab. 2025 Mar 25:S1550-4131(25)00104-4. doi: 10.1016/j.cmet.2025.03.001. Online ahead of print.

ABSTRACT

Aging affects reproductive capabilities in males through physiological and behavioral alterations, including endocrine changes and decreased libido. In this study, we investigated the influence of intermittent fasting (IF) on these aging-related declines, using male C57BL/6J mice. Our findings revealed that IF significantly preserved reproductive success in aged mice, not by improving traditional reproductive metrics such as sperm quality or endocrine functions but by enhancing mating behavior. This behavioral improvement was attributed to IF's ability to counter age-dependent increases in serotonergic inhibition, primarily through the decreased supply of the serotonin precursor tryptophan from the periphery to the brain. Our research underscores the potential of dietary interventions like IF in mitigating age-associated declines in male reproductive health and suggests a novel approach to managing conditions related to reduced sexual desire, highlighting the complex interplay between diet, metabolism, and reproductive behavior.

PMID:40157367 | DOI:10.1016/j.cmet.2025.03.001

Sci Total Environ. 2025 Mar 28;974:179243. doi: 10.1016/j.scitotenv.2025.179243. Online ahead of print.

ABSTRACT

Deforestation is one of the main drivers of environmental degradation around the world. Slash-and-burn is a common practice, performed in tropical forests to create new agricultural lands for local communities. In Madagascar, this practice affects many natural areas that host lemur habitats. Reforestation within nature reserves including fast-growing native species is desirable, for example in this area using native bamboo with the aim of restoring the habitat increased plantation success. In this context, the extensive detection of forest disturbances can effectively support restoration actions, providing an overall framework to address priorities and maximizing ecological benefits. In this work and with respect to a study area located around the Maromizaha New Protected Area (Madagascar), an analysis was conducted based on a time series of NDVI maps from Landsat missions (GSD = 30 m). The period between 1991 and 2022 was investigated to detect the location and moment of forest disturbances with the additional aim of quantifying the level of damage and of the recovery process at every disturbed location. It is worth noting that the Maromizaha New Protected Area currently hosts 12 species of endangered lemurs, highlighting its pivotal role as a critical conservation and restoration priority due to the ecological significance of preserving habitat integrity to sustain these threatened species. Detection was operated at pixel level by analyzing the local temporal profile of Normalized Difference Vegetation Index - NDVI (yearly step). Time of the eventual detected disturbance was found within the profile looking for the first derivative minimum. Significance of NDVI change was evaluated testing the Chebyshev condition and the following parameters mapped: i) year of disturbance; ii) significance of NDVI change; iii) level of damage; (iv) year of vegetation recovery; (v) rate of recovery. Accordingly, the level of the damage and the rate of recovery were used to estimate resistance and resilience indices of lemurs' habitat (inherently forested areas). Finally, temporal trends of both forest loss and recovery were analyzed to investigate potential impacts onto local lemur populations and, more in general, to the entire Reserve.

PMID:40157089 | DOI:10.1016/j.scitotenv.2025.179243

Curr Opin Biotechnol. 2025 Mar 28;93:103297. doi: 10.1016/j.copbio.2025.103297. Online ahead of print.

ABSTRACT

Oleaginous yeast are remarkably versatile organisms, distinguished by their natural capacities to accumulate high levels of neutral lipids and broad substrate range. With recent growing interests in engineering non-model organisms as superior biomanufacturing platforms, oleaginous yeasts have emerged as promising chassis for oleochemicals, terpenoids, organic acids, and other valuable products. Advancement in systems biology along with genetic tool development have significantly expanded our understanding of the metabolism in these species and enabled engineering efforts to produce biofuels and bioproducts from diverse feedstocks. This review examines the latest technical advances in oleaginous yeast research toward sustainable biomanufacturing. We cover recent developments in systems biology-enabled metabolism understanding, genetic tools, feedstock utilization, and strain engineering approaches for the production of various valuable chemicals.

PMID:40157044 | DOI:10.1016/j.copbio.2025.103297

Br J Surg. 2025 Mar 28;112(4):znaf051. doi: 10.1093/bjs/znaf051.

NO ABSTRACT

PMID:40156895 | DOI:10.1093/bjs/znaf051

BioData Min. 2025 Mar 28;18(1):27. doi: 10.1186/s13040-025-00442-z.

ABSTRACT

The integration of multi-omics data from diverse high-throughput technologies has revolutionized drug discovery. While various network-based methods have been developed to integrate multi-omics data, systematic evaluation and comparison of these methods remain challenging. This review aims to analyze network-based approaches for multi-omics integration and evaluate their applications in drug discovery. We conducted a comprehensive review of literature (2015-2024) on network-based multi-omics integration methods in drug discovery, and categorized methods into four primary types: network propagation/diffusion, similarity-based approaches, graph neural networks, and network inference models. We also discussed the applications of the methods in three scenario of drug discovery, including drug target identification, drug response prediction, and drug repurposing, and finally evaluated the performance of the methods by highlighting their advantages and limitations in specific applications. While network-based multi-omics integration has shown promise in drug discovery, challenges remain in computational scalability, data integration, and biological interpretation. Future developments should focus on incorporating temporal and spatial dynamics, improving model interpretability, and establishing standardized evaluation frameworks.

PMID:40155979 | DOI:10.1186/s13040-025-00442-z

BMC Cancer. 2025 Mar 28;25(1):560. doi: 10.1186/s12885-025-13845-2.

ABSTRACT

BACKGROUND: Colorectal cancer (CRC) remains a leading cause of cancer-related mortality, particularly due to advanced-stage metastasis. P-cadherin (CDH3), a potential therapeutic target, is highly expressed in CRC tissues and associated with poor prognosis and metastasis. However, the mechanisms underlying its role in CRC progression and its translational potential remain poorly understood.

MATERIALS AND METHODS: This study integrated multiple public databases (TCGA, HCMDB, UALCAN, HPA, UniProt, cBioPortal, and GEO) to evaluate CDH3 expression, construct a prognostic model, and perform functional and translational analyses. Immunohistochemistry was used to validate CDH3 protein expression in clinical samples. Additional analyses included correlations with clinicopathological parameters, immune infiltration (TIDE, TISIDB), functional enrichment (KEGG, GSEA), drug sensitivity (GSCA), and molecular docking (MOE). Single-cell sequencing (CancerSEA, HPA) was also conducted to explore CDH3's role at the single-cell level.

RESULTS: CDH3 expression was significantly elevated in CRC tissues and correlated with poor prognosis, recurrence, and metastasis. CDH3 expression was associated with the infiltration of resting immune cells, particularly dendritic cells, and enrichment analysis revealed its critical role in CRC metastasis through extracellular matrix (ECM) and local adhesion pathways. Notably, afatinib emerged as a promising candidate for targeting CDH3 via "drug repositioning," a process involving the repurposing of existing drugs for new therapeutic applications.

CONCLUSION: This study provides novel insights into CDH3's role in CRC metastasis and its potential as a therapeutic target. The translational potential of CDH3, including its integration with immunotherapy and drug repositioning strategies, offers a promising avenue for the treatment of metastatic CRC.

PMID:40155851 | DOI:10.1186/s12885-025-13845-2

Br J Pharmacol. 2025 Mar 28. doi: 10.1111/bph.70018. Online ahead of print.

ABSTRACT

BACKGROUND AND PURPOSE: Opioid use disorder (OUD) is a pressing public health concern marked by frequent relapse during periods of abstinence, perpetuated by negative affective states. Classical antidepressants or the currently prescribed opioid pharmacotherapies have limited efficacy to reverse the negative affect or prevent relapse.

EXPERIMENTAL APPROACH: Using mouse models, we investigated the effects of ketamine's metabolite (2R,6R)-hydroxynorketamine (HNK) on reversing conditioning to sub-effective doses of morphine in stress-susceptible mice, preventing conditioned-place aversion and alleviating acute somatic abstinence symptoms in opioid-dependent mice. Additionally, we evaluated its effects on anhedonia, anxiety-like behaviours and cognitive impairment during protracted opioid abstinence, while mechanistic studies examined cortical EEG oscillations and synaptic plasticity markers.

KEY RESULTS: (2R,6R)-HNK reversed conditioning to sub-effective doses of morphine in stress-susceptible mice and prevented conditioned-place aversion and acute somatic abstinence symptoms in opioid-dependent mice. In addition, (2R,6R)-HNK reversed anhedonia, anxiety-like behaviours and cognitive impairment emerging during protracted opioid abstinence plausibly via a restoration of impaired cortical high-frequency EEG oscillations, through a GluN2A-NMDA receptor-dependent mechanism. Notably, (2R,6R)-HNK facilitated the extinction of opioid conditioning, prevented stress-induced reinstatement of opioid-seeking behaviours and reduced the propensity for enhanced morphine self-consumption in mice previously exposed to opioids.

CONCLUSIONS AND IMPLICATIONS: These findings emphasize the therapeutic potential of (2R,6R)-HNK, which is currently in Phase II clinical trials, in addressing stress-related opioid responses. Reducing the time and cost required for development of new medications for the treatment of OUDs via drug repurposing is critical due to the opioid crisis we currently face.

PMID:40155780 | DOI:10.1111/bph.70018

Laryngoscope. 2025 Mar 29. doi: 10.1002/lary.32155. Online ahead of print.

ABSTRACT

OBJECTIVES: Olfactory dysfunction (OD) is common among people with cystic fibrosis (PwCF) and chronic rhinosinusitis (CRS). OD is associated with impaired quality of life (QOL) and dietary alterations in certain non-CF populations. This study explored relationships between OD, QOL, and modulator use in PwCF.

METHODS: This is a cross-sectional analysis of an ongoing multicenter, prospective study (2019-2023) investigating PwCF with comorbid CRS. Participants completed the 40-Question Smell Identification Test (SIT-40), 22-question SinoNasal Outcome Test-(SNOT-22), Questionnaire of Olfactory Disorders (QOD-NS), and Cystic Fibrosis Questionnaire-Revised (CFQ-R). Clinical and sinus CT data were collected. After stratification by SIT-40 score, data was analyzed by chi-square, Kruskal-Wallis, Spearman correlation, and logistic regression to identify factors associated with OD.

RESULTS: Of 59 participants, those with anosmia (n = 12) had worse eating-related QOL (CFQ-R eating) compared to individuals with normosmia (n = 16) and hyposmia (n = 31). Participants with anosmia had worse sinus CT scores than those with hyposmia. Although PwCF treated with highly effective modulator therapy (HEMT; n = 30) had better CT scores vs. non-HEMT individuals (n = 23), rates of OD in both groups were comparable. Higher SNOT-22 total scores were associated with increased odds of hyposmia or anosmia. In an eating-related QOD-NS subscore, those with worse CFQ-R eating had 2.38 times higher odds of having OD. Each point decrease in CFQ-R eating domain score was associated with 10% increased odds of OD.

CONCLUSION: In PwCF, OD was associated with increased CRS severity, impaired olfactory QOL, and decreased CFQ-R eating. There were no differences in SIT-40 or QOD-NS scores based on HEMT status.

TRIAL REGISTRATION: NCT04469439.

PMID:40156369 | DOI:10.1002/lary.32155

Int J Tuberc Lung Dis. 2025 Mar 31;29(4):153-158. doi: 10.5588/ijtld.24.0544.

ABSTRACT

<sec><title>BACKGROUND</title>Post-infectious bronchiolitis obliterans (PIBO) is a complication of severe childhood respiratory infection resulting in small airway injury, bronchiectasis, and prolonged respiratory consequences. Risk factors for PIBO and PIBO-associated bronchiectasis are unclear.</sec><sec><title>METHODS</title>This retrospective study identified all children with PIBO at a South African tertiary hospital between 1 January 2016 and 31 December 2022. The clinical characteristics, chest CT findings, and details of prior hospitalisation for respiratory infection were collected, and the characteristics of those with and without bronchiectasis were compared.</sec><sec><title>RESULTS</title>A total of 59 children were included (median age at primary lung insult: 10 months, IQR 6-17; median age at PIBO diagnosis: 16 months, IQR 11-28). Twenty-three had comorbidities, most frequently premature birth (30.5%) and HIV infection (6.8%). The most common pathogen was adenovirus (n = 41; 69.5%). At initial lung insult, 19 (32.2%) required mechanical ventilation. Mosaic attenuation on the chest CT was present in all. Thirty-three (55.9%) had bronchiectasis. The clinical characteristics, ventilation, causative pathogen, and comorbidity were similar in those with and without bronchiectasis.</sec><sec><title>CONCLUSION</title>Bronchiectasis occurs frequently in paediatric PIBO and is present within months of initial respiratory insult with no identified risk factors. Premature birth is common and may contribute to PIBO development.</sec>.

PMID:40155792 | DOI:10.5588/ijtld.24.0544

Med Phys. 2025 Mar 29. doi: 10.1002/mp.17794. Online ahead of print.

ABSTRACT

BACKGROUND: Conventional proton dose calculation methods are either time- and resource-intensive, like Monte Carlo (MC) simulations, or they sacrifice accuracy, as seen with analytical methods. This trade-off between computational efficiency and accuracy highlights the need for improved dose calculation approaches in clinical settings.

PURPOSE: This study aims to develop a deep-learning-based model that calculates dose-to-water (DW) and dose-to-medium (DM) using patient anatomy and proton spot map (PSM), achieving approaching MC-level accuracy with significantly reduced computation time. Additionally, the study seeks to generalize the model to different treatment sites using transfer learning.

METHODS: A SwinUNetr model was developed using 259 four-field prostate proton stereotactic body radiation therapy (SBRT) plans to calculate patient-specific DW and DM distributions from CT and projected PSM (PPSM). The PPSM was created by projecting PSM into the CT scans using spot coordinates, stopping power ratio, beam divergence, and water-equivalent thickness. Fine-tuning was then performed for the central nervous system (CNS) site using 84 CNS plans. The model's accuracy was evaluated against MC simulation benchmarks using mean absolute error (MAE), gamma analysis (2% local dose difference, 2-mm distance-to-agreement, 10% low dose threshold), and relevant clinical indices on the test dataset.

RESULTS: The trained model achieved a single-field dose calculation time of 0.07 s on a Nvidia-A100 GPU, over 100 times faster than MC simulators. For the prostate site, the best-performing model showed an average MAE of 0.26 ± 0.17 Gy and a gamma index of 92.2% ± 3.1% in dose regions above 10% of the maximum dose for DW calculations, and an MAE of 0.30 ± 0.19 Gy with a gamma index of 89.7% ± 3.9% for DM calculations. After transfer learning for CNS plans, the model achieved an MAE of 0.49 ± 0.24 Gy and a gamma index of 90.1% ± 2.7% for DW computations, and an MAE of 0.47 ± 0.25 Gy with a gamma index of 85.4% ± 7.1% for DM computations.

CONCLUSIONS: The SwinUNetr model provides an efficient and accurate method for computing dose distributions in proton therapy. It also opens the possibility of reverse-engineering PSM from DW, potentially speeding up treatment planning while maintaining accuracy.

PMID:40156258 | DOI:10.1002/mp.17794

Ultrason Imaging. 2025 Mar 29:1617346251319410. doi: 10.1177/01617346251319410. Online ahead of print.

ABSTRACT

Exploring the clinical significance of employing deep learning methodologies on ultrasound images for the development of an automated model to accurately identify pleomorphic adenomas and Warthin tumors in salivary glands. A retrospective study was conducted on 91 patients who underwent ultrasonography examinations between January 2016 and December 2023 and were subsequently diagnosed with pleomorphic adenoma or Warthin's tumor based on postoperative pathological findings. A total of 526 ultrasonography images were collected for analysis. Convolutional neural network (CNN) models, including ResNet18, MobileNetV3Small, and InceptionV3, were trained and validated using these images for the differentiation of pleomorphic adenoma and Warthin's tumor. Performance evaluation metrics such as receiver operating characteristic (ROC) curves, area under the curve (AUC), sensitivity, specificity, positive predictive value, and negative predictive value were utilized. Two ultrasound physicians, with varying levels of expertise, conducted independent evaluations of the ultrasound images. Subsequently, a comparative analysis was performed between the diagnostic outcomes of the ultrasound physicians and the results obtained from the best-performing model. Inter-rater agreement between routine ultrasonography interpretation by the two expert ultrasonographers and the automatic identification diagnosis of the best model in relation to pathological results was assessed using kappa tests. The deep learning models achieved favorable performance in differentiating pleomorphic adenoma from Warthin's tumor. The ResNet18, MobileNetV3Small, and InceptionV3 models exhibited diagnostic accuracies of 82.4% (AUC: 0.932), 87.0% (AUC: 0.946), and 77.8% (AUC: 0.811), respectively. Among these models, MobileNetV3Small demonstrated the highest performance. The experienced ultrasonographer achieved a diagnostic accuracy of 73.5%, with sensitivity, specificity, positive predictive value, and negative predictive value of 73.7%, 73.3%, 77.8%, and 68.8%, respectively. The less-experienced ultrasonographer achieved a diagnostic accuracy of 69.0%, with sensitivity, specificity, positive predictive value, and negative predictive value of 66.7%, 71.4%, 71.4%, and 66.7%, respectively. The kappa test revealed strong consistency between the best-performing deep learning model and postoperative pathological diagnoses (kappa value: .778, p-value < .001). In contrast, the less-experienced ultrasonographer demonstrated poor consistency in image interpretations (kappa value: .380, p-value < .05). The diagnostic accuracy of the best deep learning model was significantly higher than that of the ultrasonographers, and the experienced ultrasonographer exhibited higher diagnostic accuracy than the less-experienced one. This study demonstrates the promising performance of a deep learning-based method utilizing ultrasonography images for the differentiation of pleomorphic adenoma and Warthin's tumor. The approach reduces subjective errors, provides decision support for clinicians, and improves diagnostic consistency.

PMID:40156239 | DOI:10.1177/01617346251319410

J Forensic Sci. 2025 Mar 29. doi: 10.1111/1556-4029.70034. Online ahead of print.

ABSTRACT

With the rapid development of diffusion models, AI generation technology can now generate very realistic images. If such AI-generated images are used as evidence, they may threaten judicial fairness. Taking the adversarial identification of images generated by Craiyon V3 software as an example, this paper studies the adversarial identification methods for AI-generated image software. First, an AI-generated image set containing 18,000 images is constructed using Craiyon V3; then, an AI-generated image detection model based on deep learning is selected, and a score-based likelihood ratio method is introduced to evaluate the strength of evidence. Experimental results show that the proposed method achieves an accuracy of over 99% on multiple threshold classifiers including Swin-Transformer, ResNet-18, and so on, and the fitted likelihood ratio model also passes a series of validation criteria including Tippett plots. The research results of this paper are expected to be applied to judicial practice in the future, providing judges with a reliable and powerful decision-making basis, and laying a foundation for further exploration of AI-generated image identification methods.

PMID:40156229 | DOI:10.1111/1556-4029.70034