J Med Internet Res. 2025 Mar 24;27:e63937. doi: 10.2196/63937.

ABSTRACT

BACKGROUND: Intraoperative neurophysiological monitoring (IONM) guides the surgeon in ensuring motor pathway integrity during high-risk neurosurgical and orthopedic procedures. Although motor-evoked potentials (MEPs) are valuable for predicting motor outcomes, the key features of predictive signals are not well understood, and standardized warning criteria are lacking. Developing a muscle identification prediction model could increase patient safety while allowing the exploration of relevant features for the task.

OBJECTIVE: The aim of this study is to expand the development of machine learning (ML) methods for muscle classification and evaluate them in a bicentric setup. Further, we aim to identify key features of MEP signals that contribute to accurate muscle classification using explainable artificial intelligence (XAI) techniques.

METHODS: This study used ML and deep learning models, specifically random forest (RF) classifiers and convolutional neural networks (CNNs), to classify MEP signals from routine supratentorial neurosurgical procedures from two medical centers according to muscle identity of four muscles (extensor digitorum, abductor pollicis brevis, tibialis anterior, and abductor hallucis). The algorithms were trained and validated on a total of 36,992 MEPs from 151 surgeries in one center, and they were tested on 24,298 MEPs from 58 surgeries from the other center. Depending on the algorithm, time-series, feature-engineered, and time-frequency representations of the MEP data were used. XAI techniques, specifically Shapley Additive Explanation (SHAP) values and gradient class activation maps (Grad-CAM), were implemented to identify important signal features.

RESULTS: High classification accuracy was achieved with the RF classifier, reaching 87.9% accuracy on the validation set and 80% accuracy on the test set. The 1D- and 2D-CNNs demonstrated comparably strong performance. Our XAI findings indicate that frequency components and peak latencies are crucial for accurate MEP classification, providing insights that could inform intraoperative warning criteria.

CONCLUSIONS: This study demonstrates the effectiveness of ML techniques and the importance of XAI in enhancing trust in and reliability of artificial intelligence-driven IONM applications. Further, it may help to identify new intrinsic features of MEP signals so far overlooked in conventional warning criteria. By reducing the risk of muscle mislabeling and by providing the basis for possible new warning criteria, this study may help to increase patient safety during surgical procedures.

PMID:40127441 | DOI:10.2196/63937

PLoS One. 2025 Mar 24;20(3):e0315843. doi: 10.1371/journal.pone.0315843. eCollection 2025.

ABSTRACT

Chagas disease (CD), caused by the protozoan parasite Trypanosoma cruzi (T. cruzi), represents a major public health concern in most of the American continent and causes 12,000 deaths every year. CD clinically manifests in two phases (acute and chronic), and the diagnosis can result in complications due to the difference between phases and the long period between them. Still, strategies are lacking for the automatic diagnosis of healthy and T. cruzi-infected individuals with missing and limited data. In this work, we propose a Weighted Variational Auto-Encoder (W-VAE) for imputing and augmenting multimodal data to classify healthy individuals and individuals in the acute or chronic phases of T. cruzi infection from a murine model. W-VAE is a deep generative architecture trained with a new proposed loss function to which we added a weighting factor and a masking mechanism to improve the quality of the data generated. We imputed and augmented data using four modalities: electrocardiography signals, echocardiography images, Doppler spectrum, and ELISA antibody titers. We evaluated the generated data through different multi-classification tasks to identify healthy individuals and individuals in the acute or chronic phase of infection. In each multi-classification task, we assessed several classifiers, missing rates, and feature-selection methods. The best obtained accuracy was 92 ± 4% in training and 95% in the final test using a Gaussian Process Classifier with a missing rate of 50%. The accuracy achieved was 95% for individuals in healthy and acute phase and 100% for individuals in the chronic phase. Our approach can be useful in generating data to study the phases of T. cruzi infection.

PMID:40127396 | DOI:10.1371/journal.pone.0315843

PLoS One. 2025 Mar 24;20(3):e0309862. doi: 10.1371/journal.pone.0309862. eCollection 2025.

ABSTRACT

Citations in scientific literature act as channels for the sharing, transfer, and development of scientific knowledge. However, not all citations hold the same significance. Numerous taxonomies and machine learning models have been developed to analyze citations, but they often overlook the internal context of these citations. Moreover, it is worth noting that selecting the appropriate word embedding and classification models is crucial for achieving superior results. Word embeddings offer n-dimensional distributed representations of text, striving to capture the nuanced meanings of words. Deep learning-based word embedding techniques have garnered significant attention and found application in various Natural Language Processing (NLP) tasks, including text classification, sentiment analysis, and citation analysis. Current state-of-the-art techniques often use small datasets with fixed window sizes, resulting in the loss of contextual meaning. This study leverages two benchmark datasets encompassing a substantial volume of in-text citations to guide the selection of an optimal word embedding window size and classification approaches. A comparative analysis of various window sizes for in-text citations is conducted to identify crucial citations effectively. Additionally, Word2Vec embedding is employed in conjunction with deep learning models and machine learning models such as Convolutional Neural Networks (CNNs), Gated Recurrent Units (GRUs), Long Short-Term Memory (LSTM) networks, Support Vector Machines (SVM), Decision Trees, and Naive Bayes.The evaluation employs precision, recall, F1-score, and accuracy metrics for each combination of window sizes. The findings reveal that, particularly for lengthy in-text citations, larger citation windows are more adept at capturing the semantic essence of the references. Within the scope of this study, window sizes of 10 achieve superior accuracy and precision with both machine and deep learning models.

PMID:40127378 | DOI:10.1371/journal.pone.0309862

Br J Radiol. 2025 Mar 24:tqaf059. doi: 10.1093/bjr/tqaf059. Online ahead of print.

ABSTRACT

OBJECTIVES: To investigate the impacts of a deep learning-based iterative reconstruction algorithm on image quality and measuring accuracy of bone mineral density (BMD) in low dose chest CT.

METHODS: Phantom and patient studies were separately conducted in this study. The same low dose protocol was used for phantoms and patients. All images were reconstructed with filter back projection, hybrid iterative reconstruction (KARL, level of 3,5,7), and deep learning-based iterative reconstruction (AIIR, low, medium and high-strength). The noise power spectrum (NPS) and the task-based transfer function (TTF) were evaluated using phantom. The accuracy and the relative error (RE) of BMD were evaluated using a European spine phantom. The subjective evaluation was performed by two experienced radiologists. BMD was measured using QCT. Image noise, signal-to-noise ratio, contrast-to-noise ratio, BMD values and subjective scores were compared with Wilcoxon signed-rank test. The Cohen's kappa test was used to evaluate the inter-reader and inter-group agreement.

RESULTS: AIIR reduced noise and improved resolution in phantom images significantly. There were no significant differences among BMD values in all groups of images (all p > 0.05). RE of BMD measured with AIIR images were smaller. In objective evaluation, all strengths of AIIR achieved less image noise, higher SNR and CNR (all p < 0.05). AIIR-H showed the lowest noise, highest SNR and CNR (p < 0.05). The increase of AIIR algorithm strengths did not affect BMD values significantly (all p > 0.05).

CONCLUSION: The deep learning-based iterative reconstruction did not affect the accuracy of BMD measurement with Low-dose chest CT, while reducing image noise and improving spatial resolution.

ADVANCES IN KNOWLEDGE: The BMD values could be measured accurately in low-dose chest CT with deep learning-based iterative reconstruction, while reducing image noise and improving spatial resolution.

PMID:40127198 | DOI:10.1093/bjr/tqaf059

Brief Bioinform. 2025 Mar 4;26(2):bbaf124. doi: 10.1093/bib/bbaf124.

ABSTRACT

Accurate cancer survival prediction remains a critical challenge in clinical oncology, largely due to the complex and multi-omics nature of cancer data. Existing methods often struggle to capture the comprehensive range of informative features required for precise predictions. Here, we introduce PCLSurv, an innovative deep learning framework designed for cancer survival prediction using multi-omics data. PCLSurv integrates autoencoders to extract omics-specific features and employs sample-level contrastive learning to identify distinct yet complementary characteristics across data views. Then, features are fused via a bilinear fusion module to construct a unified representation. To further enhance the model's capacity to capture high-level semantic relationships, PCLSurv aligns similar samples with shared prototypes while separating unrelated ones via prototypical contrastive learning. As a result, PCLSurv effectively distinguishes patient groups with varying survival outcomes at different semantic similarity levels, providing a robust framework for stratifying patients based on clinical and molecular features. We conduct extensive experiments on 11 cancer datasets. The comparison results confirm the superior performance of PCLSurv over existing alternatives. The source code of PCLSurv is freely available at https://github.com/LiangSDNULab/PCLSurv.

PMID:40127182 | DOI:10.1093/bib/bbaf124

Brief Bioinform. 2025 Mar 4;26(2):bbaf122. doi: 10.1093/bib/bbaf122.

ABSTRACT

As part of the drug repurposing process, it is imperative to predict the interactions between drugs and target proteins in an accurate and efficient manner. With the introduction of contrastive learning into drug-target prediction, the accuracy of drug repurposing will be further improved. However, a large part of DTI prediction methods based on deep learning either focus only on the structural features of proteins and drugs extracted using GNN or CNN, or focus only on their relational features extracted using heterogeneous graph neural networks on a DTI heterogeneous graph. Since the structural and relational features of proteins and drugs describe their attribute information from different perspectives, their combination can improve DTI prediction performance. We propose a relational similarity-based graph contrastive learning for DTI prediction (RSGCL-DTI), which combines the structural and relational features of drugs and proteins to enhance the accuracy of DTI predictions. In our proposed method, the inter-protein relational features and inter-drug relational features are extracted from the heterogeneous drug-protein interaction network through graph contrastive learning, respectively. The results demonstrate that combining the relational features obtained by graph contrastive learning with the structural ones extracted by D-MPNN and CNN enhances feature representation ability, thereby improving DTI prediction performance. Our proposed RSGCL-DTI outperforms eight SOTA baseline models on the four benchmark datasets, performs well on the imbalanced dataset, and also shows excellent generalization ability on unseen drug-protein pairs.

PMID:40127181 | DOI:10.1093/bib/bbaf122

Brief Bioinform. 2025 Mar 4;26(2):bbaf119. doi: 10.1093/bib/bbaf119.

ABSTRACT

Accurately predicting the pathogenicity of missense variants is crucial for improving disease diagnosis and advancing clinical research. However, existing computational methods primarily focus on general pathogenicity predictions, overlooking assessments of disease-specific conditions. In this study, we propose DS-MVP, a method capable of predicting disease-specific pathogenicity of missense variants in human genomes. DS-MVP first leverages a deep learning model pre-trained on a large general pathogenicity dataset to learn rich representation of missense variants. It then fine-tunes these representations with an XGBoost model on smaller datasets for specific diseases. We evaluated the learned representation by testing it on multiple binary pathogenicity datasets and gene-level statistics, demonstrating that DS-MVP outperforms existing state-of-the-art methods, such as MetaRNN and AlphaMissense. Additionally, DS-MVP excels in multi-label and multi-class classification, effectively classifying disease-specific pathogenic missense variants based on disease conditions. It further enhances predictions by fine-tuning the pre-trained model on disease-specific datasets. Finally, we analyzed the contributions of the pre-trained model and various feature types, with gene description corpus features from large language model and genetic feature fusion contributing the most. These results underscore that DS-MVP represents a broader perspective on pathogenicity prediction and holds potential as an effective tool for disease diagnosis.

PMID:40127180 | DOI:10.1093/bib/bbaf119

J Chem Inf Model. 2025 Mar 24. doi: 10.1021/acs.jcim.4c02291. Online ahead of print.

ABSTRACT

The potential of hydrogen (pH) influences the function of the enzyme. Measuring or predicting the optimal pH (pHopt) at which enzymes exhibit maximal catalytic activity is crucial for enzyme design and application. The rapid development of enzyme mining and de novo design has produced a large number of new enzymes, making it impractical to measure their pHopt in the wet laboratory. Consequently, in-silico computational approaches such as machine learning and deep learning models, which offer pH prediction at minimal cost, have attracted considerable interest. This work presents Venus-DREAM, an enzyme pHopt prediction model based on the kNN algorithm and few-shot learning, which achieves state-of-the-art accuracy in pHopt prediction. Venus-DREAM regards the pHopt prediction of an enzyme as a few-shot learning task: learning from the k-closest labeled enzymes to predict the pHopt of the target enzyme. The value of k is determined by the optimal k-value of the kNN regression algorithm. And the distance between two enzymes is defined as the cosine similarity of their mean-pooled embeddings obtained from protein language models (PLMs). The few-shot learner is based on the Reptile algorithm, which first adapts to the k-nearest labeled enzymes to create a specialized model for the target enzyme and then predicts its pHopt. This efficient method enables high-throughput virtual exploration of protein space, facilitating the identification of sequences with the desired pHopt ranges in a high-throughput manner. Moreover, our method can be easily adapted in other protein function prediction tasks.

PMID:40127128 | DOI:10.1021/acs.jcim.4c02291

IEEE Trans Image Process. 2025 Mar 24;PP. doi: 10.1109/TIP.2025.3552198. Online ahead of print.

ABSTRACT

Ultrasound Localization Microscopy (ULM) is a non-invasive technique that allows for the imaging of micro-vessels in vivo, at depth and with a resolution on the order of ten microns. ULM is based on the sub-resolution localization of individual microbubbles injected in the bloodstream. Mapping the whole angioarchitecture requires the accumulation of microbubbles trajectories from thousands of frames, typically acquired over a few minutes. ULM acquisition times can be reduced by increasing the microbubble concentration, but requires more advanced algorithms to detect them individually. Several deep learning approaches have been proposed for this task, but they remain limited to 2D imaging, in part due to the associated large memory requirements. Herein, we propose the use of sparse tensor neural networks to enable deep learning-based 3D ULM by improving memory scalability with increased dimensionality. We study several approaches to efficiently convert ultrasound data into a sparse format and study the impact of the associated loss of information. When applied in 2D, the sparse formulation reduces the memory requirements by a factor 2 at the cost of a small reduction of performance when compared against dense networks. In 3D, the proposed approach reduces memory requirements by two order of magnitude while largely outperforming conventional ULM in high concentration settings. We show that Sparse Tensor Neural Networks in 3D ULM allow for the same benefits as dense deep learning based method in 2D ULM i.e. the use of higher concentration in silico and reduced acquisition time.

PMID:40126968 | DOI:10.1109/TIP.2025.3552198

IEEE J Biomed Health Inform. 2025 Mar 24;PP. doi: 10.1109/JBHI.2025.3553502. Online ahead of print.

ABSTRACT

Depression is one of the most common mood disorders and the number of patients increases significantly in recent years. Due to the lack of biomarkers, conversation between patients and psychiatrists is still the main clinical diagnostic method which is easily influenced by subjectivity of both patients and psychiatrists. Synthetic House-tree-person test (S-HTP), a convenient and efficient mental assessment tool, minimizes subjective influences from patients, while its effectiveness is limited by the professional ability of analyst. Here we introduce a deep learning model DeHTP, a flexible and convenient depression detection method based on S-HTP without interaction between people. Experimental results demonstrate that DeHTP achieves 0.963 AUC and 0.9 accuracy, and outperforms the conventional manual analysis of S-HTP, which is conducted on the guideline of 50 conclusions from previous study related to depression. In addition, it reveals 22 depression-correlated drawing features aligned with conclusions above from the perspective of our proposed model. Leveraging the advantages of deep learning and S-HTP, this approach has the potential for widespread promotion and adoption as the available tool for daily self-mental monitoring, as well as the promising auxiliary diagnostic method in clinical.

PMID:40126963 | DOI:10.1109/JBHI.2025.3553502

ACS Sens. 2025 Mar 24. doi: 10.1021/acssensors.4c02789. Online ahead of print.

ABSTRACT

Gas sensor arrays designed for pattern recognition face persistent challenges in achieving high sensitivity and selectivity for multiple volatile organic compounds (VOCs), particularly under varying environmental conditions. To address these limitations, we developed multimodal intelligent MEMS gas sensors by precisely tailoring the nanocomposite ratio of NiO and ZnO components. These sensors demonstrate enhanced responses to ethylene glycol (EG) and limonene (LM) at different operating temperatures, demonstrating material-specific selectivity. Additionally, a multitask deep learning model is employed for real-time, quantitative detection of VOCs, accurately predicting their concentration and type. These results showcase the effectiveness of combining material optimization with advanced algorithms for real-world VOCs detection, advancing the field of odor analysis tools.

PMID:40126565 | DOI:10.1021/acssensors.4c02789

J Med Internet Res. 2025 Mar 24;27:e67922. doi: 10.2196/67922.

ABSTRACT

BACKGROUND: Emerging evidence underscores the potential application of artificial intelligence (AI) in discovering noninvasive blood biomarkers. However, the diagnostic value of AI-derived blood biomarkers for ovarian cancer (OC) remains inconsistent.

OBJECTIVE: We aimed to evaluate the research quality and the validity of AI-based blood biomarkers in OC diagnosis.

METHODS: A systematic search was performed in the MEDLINE, Embase, IEEE Xplore, PubMed, Web of Science, and the Cochrane Library databases. Studies examining the diagnostic accuracy of AI in discovering OC blood biomarkers were identified. The risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies-AI tool. Pooled sensitivity, specificity, and area under the curve (AUC) were estimated using a bivariate model for the diagnostic meta-analysis.

RESULTS: A total of 40 studies were ultimately included. Most (n=31, 78%) included studies were evaluated as low risk of bias. Overall, the pooled sensitivity, specificity, and AUC were 85% (95% CI 83%-87%), 91% (95% CI 90%-92%), and 0.95 (95% CI 0.92-0.96), respectively. For contingency tables with the highest accuracy, the pooled sensitivity, specificity, and AUC were 95% (95% CI 90%-97%), 97% (95% CI 95%-98%), and 0.99 (95% CI 0.98-1.00), respectively. Stratification by AI algorithms revealed higher sensitivity and specificity in studies using machine learning (sensitivity=85% and specificity=92%) compared to those using deep learning (sensitivity=77% and specificity=85%). In addition, studies using serum reported substantially higher sensitivity (94%) and specificity (96%) than those using plasma (sensitivity=83% and specificity=91%). Stratification by external validation demonstrated significantly higher specificity in studies with external validation (specificity=94%) compared to those without external validation (specificity=89%), while the reverse was observed for sensitivity (74% vs 90%). No publication bias was detected in this meta-analysis.

CONCLUSIONS: AI algorithms demonstrate satisfactory performance in the diagnosis of OC using blood biomarkers and are anticipated to become an effective diagnostic modality in the future, potentially avoiding unnecessary surgeries. Future research is warranted to incorporate external validation into AI diagnostic models, as well as to prioritize the adoption of deep learning methodologies.

TRIAL REGISTRATION: PROSPERO CRD42023481232; https://www.crd.york.ac.uk/PROSPERO/view/CRD42023481232.

PMID:40126546 | DOI:10.2196/67922

Invest Ophthalmol Vis Sci. 2025 Mar 3;66(3):49. doi: 10.1167/iovs.66.3.49.

ABSTRACT

PURPOSE: Optic nerve crush (ONC) is a model for studying optic nerve trauma. Unilateral ONC induces massive retinal ganglion cell (RGC) degeneration in the affected eye, leading to vision loss within a month. A common assumption has been that the non-injured contralateral eye is unaffected due to the minimal retino-retinal projections of the RGCs at the chiasm. Yet, recently, microglia, the brain-resident macrophages, have shown a responsive phenotype in the contralateral eye after ONC. Whether RGC loss accompanies this phenotype is still controversial.

METHODS: Using the available RGCode algorithm and developing our own RGC-Quant deep-learning-based tool, we quantify RGC's total number and density across the entire retina after ONC.

RESULTS: We confirm a short-term microglia response in the contralateral eye after ONC, but this did not affect the microglia number. Furthermore, we cannot confirm the previously reported RGC loss between naïve and contralateral retinas 5 weeks after ONC induction across the commonly used Cx3cr1creERT2 and C57BL6/J mouse models. Neither sex nor the direct comparison of the RGC markers Brn3a and RBPMS, with Brn3a co-labeling, on average, 89% of the RBPMS+-cells, explained this discrepancy, suggesting that the early microglia-responsive phenotype does not have immediate consequences on the RGC number.

CONCLUSIONS: Our results corroborate that unilateral optic nerve injury elicits a microglial response in the uninjured contralateral eye but without RGC loss. Therefore, the contralateral eye should be treated separately and not as an ONC control.

PMID:40126507 | DOI:10.1167/iovs.66.3.49

Noncoding RNA. 2025 Mar 7;11(2):23. doi: 10.3390/ncrna11020023.

ABSTRACT

(1) Background: MicroRNAs are non-coding RNA sequences that regulate cellular functions by targeting messenger RNAs and inhibiting protein synthesis. Identifying their target sites is vital to understanding their roles. However, it is challenging due to the high cost and time demands of experimental methods and the high false-positive rates of computational approaches. (2) Methods: We introduce a Multi-Input Neural Network (MINN) algorithm that integrates diverse biologically relevant features, including the microRNA duplex structure, substructures, minimum free energy, and base-pairing probabilities. For each feature derived from a microRNA target-site duplex, we create a corresponding image. These images are processed in parallel by the MINN algorithm, allowing it to learn a comprehensive and precise representation of the underlying biological mechanisms. (3) Results: Our method, on an experimentally validated test set, detects target sites with an AUPRC of 0.9373, Precision of 0.8725, and Recall of 0.8703 and outperforms several commonly used computational methods of microRNA target-site predictions. (4) Conclusions: Incorporating diverse biologically explainable features, such as duplex structure, substructures, their MFEs, and binding probabilities, enables our model to perform well on experimentally validated test data. These features, rather than nucleotide sequences, enhance our model to generalize beyond specific sequence contexts and perform well on sequentially distant samples.

PMID:40126347 | DOI:10.3390/ncrna11020023

Noncoding RNA. 2025 Feb 26;11(2):18. doi: 10.3390/ncrna11020018.

ABSTRACT

RNA inverse design is an essential part of many RNA therapeutic strategies. To date, there have been great advances in computationally driven RNA design. The current machine learning approaches can predict the sequence of an RNA given its 3D structure with acceptable accuracy and at tremendous speed. The design and engineering of RNA regulators such as riboswitches, however, is often more difficult, partly due to their inherent conformational switching abilities. Although recent state-of-the-art models do incorporate information about the multiple structures that a sequence can fold into, there is great room for improvement in modeling structural switching. In this work, a relational geometric graph neural network is proposed that explicitly incorporates alternative structures to predict an RNA sequence. Converting the RNA structure into a geometric graph, the proposed model uses edge types to distinguish between the primary structure, secondary structure, and spatial positioning of the nucleotides in representing structures. The results show higher native sequence recovery rates over those of gRNAde across different test sets (eg. 72% vs. 66%) and a benchmark from the literature (60% vs. 57%). Secondary-structure edge types had a more significant impact on the sequence recovery than the spatial edge types as defined in this work. Overall, these results suggest the need for more complex and case-specific characterization of RNA for successful inverse design.

PMID:40126342 | DOI:10.3390/ncrna11020018

Soft Matter. 2025 Mar 24. doi: 10.1039/d4sm01117a. Online ahead of print.

ABSTRACT

Bacteria aggregate through various intercellular interactions to build biofilms, but the effect of environmental changes on them remains largely unexplored. Here, by using an experimental device that overcomes past difficulties, we observed the collective response of Escherichia coli aggregates to dynamic changes in the growth conditions. We discovered that nutrient starvation caused bacterial cells to arrange themselves into bundle-shaped clusters, developing a structure akin to that of smectic liquid crystals. The degree of the smectic-like bundle order was evaluated by a deep learning approach. Our experiments suggest that both the depletion attraction by extracellular polymeric substances and the growth arrest are essential for the bundle formation. Since these effects of nutrient starvation at the single-cell level are common to many bacterial species, bundle formation might also be a common collective behavior that bacterial cells may exhibit under harsh environments.

PMID:40126189 | DOI:10.1039/d4sm01117a

Adv Sci (Weinh). 2025 Mar 24:e2502744. doi: 10.1002/advs.202502744. Online ahead of print.

ABSTRACT

The liver harbors complex cross-scale structures, and the fibrosis-related alterations to these structures have a severe impact on the diverse function of the liver. However, the hepatic anatomic structures and their pathological alterations in the whole-liver scale remain to be elucidated. Combining the micro-optical sectioning tomography (MOST) system and liver Nissl staining, a first high-precision whole mouse liver atlas is generated, enabling visualization and analysis of the entire mouse liver. Thus, a detailed 3D panorama of CCl4-induced liver fibrosis pathology is constructed, capturing the 3D details of the central veins, portal veins, arteries, bile ducts, hepatic sinusoids, and liver cells. Pathological changes, including damaged sinusoids, steatotic hepatocytes, and collagen deposition, are region-specific and concentrated in the pericentral areas. The quantitative analysis shows a significantly reduced diameter and increased length density of the central vein. Additionally, a deep learning tool is used to segment steatotic hepatocytes, finding that the volume proportion of steatotic regions is similar across liver lobes. Steatosis severity increases with proximity to the central vein, independent of central vein diameter. The approach allows the cross-scale visualization of multiple structural components in liver research and promotes pathological studies from a 2D to a 3D perspective.

PMID:40126158 | DOI:10.1002/advs.202502744

Network. 2025 Mar 24:1-37. doi: 10.1080/0954898X.2025.2472626. Online ahead of print.

ABSTRACT

Plant diseases significantly threaten food security by reducing the quantity and quality of agricultural products. This paper presents a deep learning approach for classifying foliar diseases in apple plants using the Tunicate Swarm Sine Cosine Algorithm-based Deep Residual Network (TSSCA-based DRN). Cluster heads in simulated Internet of Things (IoT) networks are selected by Fractional Lion Optimization (FLION), and images are pre-processed with a Gaussian filter and segmented using the DeepJoint model. The TSSCA, combining the Tunicate Swarm Algorithm (TSA) and Sine Cosine Algorithm (SCA), enhances the classifier's effectiveness. Moreover, Plant Pathology 2020 - FGVC7 dataset is used in this work. This dataset is designed for the classification of foliar diseases in apple trees. The TSSCA-based DRN outperforms other methods, achieving 97% accuracy, 94.666% specificity, 96.888% sensitivity, and 0.0442J maximal energy, with significant improvements over existing approaches. Additionally, the proposed model demonstrates superior accuracy, outperforming other methods by 8.97%, 6.58%, 2.07%, 1.71%, 1.14%, 1.07%, 0.93%, and 0.64% over Multidimensional Feature Compensation Residual neural network (MDFC - ResNet), Convolutional Neural Network (CNN), Multi-Context Fusion Network (MCFN), Advanced Segmented Dimension Extraction (ASDE), and DRN, fuzzy deep convolutional neural network (FCDCNN), ResNet9-SE, Capsule Neural Network (CapsNet), IoT-based scrutinizing model, and Multi-Model Fusion Network (MMF-Net).

PMID:40126079 | DOI:10.1080/0954898X.2025.2472626

Network. 2025 Mar 24:1-26. doi: 10.1080/0954898X.2025.2480294. Online ahead of print.

ABSTRACT

This study aims to enhance the educational security and legitimacy by overcoming the problem of real-time student signature verification. The issue is raised from the growing issue about identity theft and academic fraud in schools, which compromises the validity of tests and other academic evaluations. To overcome these problems, the paper presents a deep learning-based method for signature verification made possible by employing the cutting-edge Convolutional Neural Networks (CNNs). The proposed method utilizes a VGG19 architecture trained and adjusted to handle the unique characteristics of student signatures. Initially, the procedure is pre-processing the image, after the key signature features are extracted. After passing these characteristics across VGG19 network, the signature's authenticity is classified as either unreliable or malicious nodes. The proposed method offers a flexibility and scalability for various educational settings with its capacity to manage both batch and individual processing. The model's efficacy is demonstrated by experiment with accuracy, precision, and recall values, which surpasses the existing techniques. The method ensures dependable performance under circumstances by illustrating resilience to several kinds of noise and distortion. The proposed deep learning model results pay a way for addressing the issue of student signature verification, enhancing the academic institutions' security and legitimacy.

PMID:40126006 | DOI:10.1080/0954898X.2025.2480294

Network. 2025 Mar 24:1-41. doi: 10.1080/0954898X.2025.2480299. Online ahead of print.

ABSTRACT

Breast cancer is the foremost cause of mortality among females. Early diagnosis of a disease is necessary to avoid breast cancer by reducing the death rate and offering a better life to the individuals. Therefore, this work proposes a Parallel Convolutional SpinalNet (PConv-SpinalNet) for the efficient detection of breast cancer using mammogram images. At first, the input image is pre-processed using the Gabor filter. The tumour segmentation is conducted using LadderNet. Then, the segmented tumour samples are augmented using Image manipulation, Image erasing, and Image mix techniques. After that, the essential features, like CNN features, Texton, Local Gabor binary patterns (LGBP), scale-invariant feature transform (SIFT), and Local Monotonic Pattern (LMP) with discrete cosine transform (DCT) are extracted in the feature extraction phase. Finally, the detection of breast cancer is performed using PConv-SpinalNet. PConv-SpinalNet is developed by an integration of Parallel Convolutional Neural Networks (PCNN) and SpinalNet. The evaluation results show that PConv-SpinalNet accomplished a superior range of accuracy as 88.5%, True Positive Rate (TPR) as 89.7%, True Negative Rate (TNR) as 90.7%, Positive Predictive Value (PPV) as 91.3%, and Negative Predictive Value (NPV) as 92.5%.

PMID:40125951 | DOI:10.1080/0954898X.2025.2480299