Neuroradiology. 2025 May 15. doi: 10.1007/s00234-025-03631-z. Online ahead of print.

ABSTRACT

PURPOSE: We conducted a systematic review and meta-analysis to evaluate the performance of magnetic resonance imaging (MRI)-derived deep learning (DL) models in predicting 1p/19q codeletion status in glioma patients.

METHODS: The literature search was performed in four databases: PubMed, Web of Science, Embase, and Scopus. We included the studies that evaluated the performance of end-to-end DL models in predicting the status of glioma 1p/19q codeletion. The quality of the included studies was assessed by the Quality assessment of diagnostic accuracy studies-2 (QUADAS-2) METhodological RadiomICs Score (METRICS). We calculated diagnostic pooled estimates and heterogeneity was evaluated using I2. Subgroup analysis and sensitivity analysis were conducted to explore sources of heterogeneity. Publication bias was evaluated by Deeks' funnel plots.

RESULTS: Twenty studies were included in the systematic review. Only two studies had a low quality. A meta-analysis of the ten studies demonstrated a pooled sensitivity of 0.77 (95% CI: 0.63-0.87), a specificity of 0.85 (95% CI: 0.74-0.92), a positive diagnostic likelihood ratio (DLR) of 5.34 (95% CI: 2.88-9.89), a negative DLR of 0.26 (95% CI: 0.16-0.45), a diagnostic odds ratio of 20.24 (95% CI: 8.19-50.02), and an area under the curve of 0.89 (95% CI: 0.86-0.91). The subgroup analysis identified a significant difference between groups depending on the segmentation method used.

CONCLUSION: DL models can predict glioma 1p/19q codeletion status with high accuracy and may enhance non-invasive tumor characterization and aid in the selection of optimal therapeutic strategies.

PMID:40369298 | DOI:10.1007/s00234-025-03631-z

NPJ Digit Med. 2025 May 14;8(1):277. doi: 10.1038/s41746-025-01577-3.

ABSTRACT

Neuropathic corneal pain (NCP) is an underdiagnosed ocular disorder caused by aberrant nociception and hypersensitivity of corneal nerves, often resulting in chronic pain and discomfort even in the absence of noxious stimuli. Recently, microneuromas (aberrant growth and swelling of the corneal nerve endings) detected using in vivo confocal microscopy (IVCM) have emerged as a promising biomarker for NCP. However, this process is time-intensive and error-prone, limiting its clinical use and availability. In this work, we present a new NCP screening system based on a deep learning model trained to detect microneuromas using a multisite dataset with a combined total of 103,168 IVCM images. Our model showed excellent discriminative ability detecting microneuromas (AuROC: 0.97) and the ability to generalize to data from a new institution (AuROC: 0.90). Additionally, our pipeline provides an uncertainty quantification mechanism that allows it to communicate when its predictions are reliable, further increasing its clinical relevance.

PMID:40369269 | DOI:10.1038/s41746-025-01577-3

Clin Exp Metastasis. 2025 May 15;42(3):30. doi: 10.1007/s10585-025-10349-y.

ABSTRACT

Local recurrence and distant metastasis were a common manifestation of locoregionally advanced nasopharyngeal carcinoma (LA-NPC) after neoadjuvant chemoradiotherapy (NACT). To validate the clinical value of MRI radiomic models based on deep learning for predicting the recurrence of LA-NPC patients. A total of 328 NPC patients from four hospitals were retrospectively included and divided into the training(n = 229) and validation (n = 99) cohorts randomly. Extracting 975 traditional radiomic features and 1000 deep radiomic features from contrast enhanced T1-weighted (T1WI + C) and T2-weighted (T2WI) sequences, respectively. Least absolute shrinkage and selection operator (LASSO) was applied for feature selection. Five machine learning classifiers were conducted to develop three models for LA-NPC prediction in training cohort, namely Model I: traditional radiomic features, Model II: combined the deep radiomic features with Model I, and Model III: combined Model II with clinical features. The predictive performance of these models were evaluated by receive operating characteristic (ROC) curve analysis, area under the curve (AUC), accuracy, sensitivity and specificity in both cohorts. The clinical characteristics in two cohorts showed no significant differences. Choosing 15 radiomic features and 6 deep radiomic features from T1WI + C. Choosing 9 radiomic features and 6 deep radiomic features from T2WI. In T2WI, the Model II based on Random forest (RF) (AUC = 0.87) performed best compared with other models in validation cohort. Traditional radiomic model combined with deep radiomic features shows excellent predictive performance. It could be used assist clinical doctors to predict curative effect for LA-NPC patients after NACT.

PMID:40369240 | DOI:10.1007/s10585-025-10349-y

Sci Rep. 2025 May 14;15(1):16815. doi: 10.1038/s41598-025-99460-4.

ABSTRACT

Knee osteoarthritis (KOA) is a severe arthrodial joint condition with significant global socioeconomic consequences. Early recognition and treatment of KOA is critical for avoiding disease progression and developing effective treatment programs. The prevailing method for knee joint analysis involves manual diagnosis, segmentation, and annotation to diagnose osteoarthritis (OA) in clinical practice while being highly laborious and a susceptible variable among users. To address the constraints of this method, several deep learning techniques, particularly the deep convolutional neural networks (CNNs), were applied to increase the efficiency of the proposed workflow. The main objective of this study is to create advanced deep learning (DL) approaches for risk assessment to forecast the evolution of pain for people suffering from KOA or those at risk of developing it. The suggested methodology applies a collective transfer learning approach for extracting accurate deep features using four pre-trained models, VGG19, ResNet50, AlexNet, and GoogleNet, to extract features from KOA images. The numeral of extracted features was reduced for identifying the most appropriate feature attributes for the disease. The binary Greylag Goose (bGGO) optimizer was employed to perform this task, with an average fitness of 0.4137 and a best fitness of 0.3155. The chosen features were categorized utilizing both deep learning and machine learning approaches. Finally, a CNN hyper-parameter algorithm was performed utilizing GGO. The suggested model outperformed previous models with accuracy, sensitivity, and specificity of 0.988692, 0.980156, and 0.990089, respectively. A comprehensive statistical analysis test was performed to confirm the validity of our findings.

PMID:40369219 | DOI:10.1038/s41598-025-99460-4

Sci Rep. 2025 May 14;15(1):16768. doi: 10.1038/s41598-025-91212-8.

ABSTRACT

Underwater computer vision faces significant challenges from light scattering, absorption, and poor illumination, which severely impact underwater vision tasks. To address these issues, ViT-Clarity, an underwater image enhancement module, is introduced, which integrates vision transformers with a convolutional neural network for superior performance. For comparison, ClarityNet, a transformer-free variant of the architecture, is presented to highlight the transformer's impact. Given the limited availability of paired underwater image datasets (clear and degraded), BlueStyleGAN is proposed as a generative model to create synthetic underwater images from clear in-air images by simulating realistic attenuation effects. BlueStyleGAN is evaluated against existing state-of-the-art synthetic dataset generators in terms of training stability and realism. Vit-ClarityNet is rigorously tested on five datasets representing diverse underwater conditions and compared with recent state-of-the-art methods as well as ClarityNet. Evaluations include qualitative and quantitative metrics such as UCIQM, UCIQE, and the deep learning-based URanker. Additionally, the impact of enhanced images on object detection and SIFT feature matching is assessed, demonstrating the practical benefits of image enhancement for underwater computer vision tasks.

PMID:40369132 | DOI:10.1038/s41598-025-91212-8

Sci Rep. 2025 May 14;15(1):16681. doi: 10.1038/s41598-025-01841-2.

ABSTRACT

In Wireless Sensor Networks (WSNs), achieving optimal coverage in dynamic environments remains a significant challenge. Traditional optimization techniques, such as genetic algorithms, particle swarm optimization, and ant colony optimization, have demonstrated adaptability in node placement but struggle with real-time self-learning capabilities, requiring frequent retraining to handle continuously changing conditions. To address these limitations, this research introduces a novel hybrid model that integrates Deep Reinforcement Learning (DRL) with Graph Neural Networks (GNN). The DRL component enables adaptive decision-making, allowing real-time sensor node adjustments based on network performance feedback. Simultaneously, the GNN model enhances spatial awareness by capturing relational dependencies among sensor nodes, optimizing coverage efficiency. This integration significantly improves network adaptability and operational efficiency. Extensive simulations demonstrate that the proposed DRL-GNN model achieves a coverage ratio of up to 96.4%, energy efficiency of 95.8%, and minimizes overlap to 5.2%, outperforming traditional methods. These results validate the effectiveness of the proposed approach in enhancing WSN coverage while maintaining energy efficiency and minimal redundancy.

PMID:40369115 | DOI:10.1038/s41598-025-01841-2

Comput Biol Med. 2025 May 13;192(Pt B):110304. doi: 10.1016/j.compbiomed.2025.110304. Online ahead of print.

ABSTRACT

Percutaneous screw fixation technique in pelvic trauma surgery is an extremely challenging operation that typically requires a trial-and-error insertion process under the guidance of continuous intraoperative X-ray. This process can be simplified by utilizing surgical navigation systems. Understanding the complexity of the intraosseous pelvis corridor is essential for establishing the optimal screw corridor, which further facilitates preoperative planning and intraoperative application. Traditional screw corridor search algorithms necessitate traversing the entrance and exit areas of the screw and calculating the distance from the corridor axis to the bone surface to ascertain the location of the screw. This process is computationally complex, and manual measurement by the physician is time consuming, labor intensive, and empirically dependent. In this study, we propose an automated planning algorithm for pelvic screw corridors based on deep learning deformable registration technology, which can efficiently and accurately identify the optimal screw corridors. Compared to traditional methods, the innovations of this study include: (1) the introduction of corridor safety range constraints on screw positioning, which enhances search efficiency; (2) the application of deep learning deformable registration to facilitate the automatic annotation of the screw entrance and exit areas, as well as the safety range of the corridor; and (3) the development of a highly efficient algorithm for optimal corridor searching, quickly determining the corridor without traversing the entrance and exit areas and enhancing efficiency via a vector-based diameter calculation method. The whole framework of the algorithm consists of three key components: atlas generation module, deformable registration and optimal corridor searching strategy. In the experiments, we test the performance of the proposed algorithm on 198 intact pelvises for calculating the optimal corridor of anterior column corridor and S1 sacroiliac screws. The results show that the new algorithm can increase the corridor diameter by 2.1%-3.3% compared to manual measurements, while significantly reducing the average time from 1038s and 3398s to 18.9s and 26.7s on anterior column corridor and S1 sacroiliac corridor, respectively, compared to the traditional screw searching algorithm. This demonstrates the advantages of the algorithm in terms of efficiency and accuracy. However, the current method is validated only on intact pelvises; further research is required for pelvic fracture scenarios.

PMID:40367630 | DOI:10.1016/j.compbiomed.2025.110304

Comput Biol Med. 2025 May 13;192(Pt B):110266. doi: 10.1016/j.compbiomed.2025.110266. Online ahead of print.

ABSTRACT

Accurate and secure classifying informal documents related to mental disorders is challenging due to factors such as informal language, noisy data, cultural differences, personal information and mixed emotions. Conventional deep learning models often struggle to capture patterns in informal text, as they miss long-range dependencies, explain words and phrases literally, and have difficulty processing non-standard inputs like emojis. To address these limitations, we expand data dimensionality, transforming and fusing textual data and signs from a 1D to a 2D space. This enables the use of pre-trained 2D CNN models, such as AlexNet, Restnet-50, and VGG-16 removing the need to design and train new models from scratch. We apply this approach to a dataset of social media posts to classify informal documents as either related to suicide or non-suicide content. Our results demonstrate high classification accuracy, exceeding 99%. In addition, our 2D visual data representation conceals individual private information and helps explainability.

PMID:40367624 | DOI:10.1016/j.compbiomed.2025.110266

Eur J Radiol. 2025 Apr 27;188:112137. doi: 10.1016/j.ejrad.2025.112137. Online ahead of print.

ABSTRACT

Accurate segmentation of cerebral arteries on computed tomography angiography (CTA) images is essential for the diagnosis and management of cerebrovascular diseases, including ischemic stroke. This study implemented a deep learning-based U-Net++ model for cerebral artery segmentation in CTA images, focusing on optimizing pruning levels by analyzing the trade-off between segmentation performance and computational cost. Dual-energy CTA and direct subtraction CTA datasets were utilized to segment the internal carotid and vertebral arteries in close proximity to the bone. We implemented four pruning levels (L1-L4) in the U-Net++ model and evaluated the segmentation performance using accuracy, intersection over union, F1-score, boundary F1-score, and Hausdorff distance. Statistical analyses were conducted to assess the significance of segmentation performance differences across pruning levels. In addition, we measured training and inference times to evaluate the trade-off between segmentation performance and computational efficiency. Applying deep supervision improved segmentation performance across all factors. While the L4 pruning level achieved the highest segmentation performance, L3 significantly reduced training and inference times (by an average of 51.56 % and 22.62 %, respectively), while incurring only a small decrease in segmentation performance (7.08 %) compared to L4. These results suggest that L3 achieves an optimal balance between performance and computational cost. This study demonstrates that pruning levels in U-Net++ models can be optimized to reduce computational cost while maintaining effective segmentation performance. By simplifying deep learning models, this approach can improve the efficiency of cerebrovascular segmentation, contributing to faster and more accurate diagnoses in clinical settings.

PMID:40367559 | DOI:10.1016/j.ejrad.2025.112137

Comput Methods Programs Biomed. 2025 May 10;268:108843. doi: 10.1016/j.cmpb.2025.108843. Online ahead of print.

ABSTRACT

BACKGROUND: Three-dimensional (3D) ultrasound (US) reconstruction is of significant value in clinical diagnosis, characterized by its safety, portability, low cost, and high real-time capabilities. 3D freehand ultrasound reconstruction aims to eliminate the need for tracking devices, relying solely on image data to infer the spatial relationships between frames. However, inherent jitter during handheld scanning introduces significant inaccuracies, making current methods ineffective in precisely predicting the spatial motions of ultrasound image frames. This leads to substantial cumulative errors over long sequence modeling, resulting in deformations or artifacts in the reconstructed volume. To address these challenges, we proposed UltrasOM, a 3D ultrasound reconstruction network designed for spatial relative motion estimation.

METHODS: Initially, we designed a video embedding module that integrates optical flow dynamics with original static information to enhance motion change features between frames. Next, we developed a Mamba-based spatiotemporal attention module, utilizing multi-layer stacked Space-Time Blocks to effectively capture global spatiotemporal correlations within video frame sequences. Finally, we incorporated correlation loss and motion speed loss to prevent overfitting related to scanning speed and pose, enhancing the model's generalization capability.

RESULTS: Experimental results on a dataset of 200 forearm cases, comprising 58,011 frames, demonstrated that the proposed method achieved a final drift rate (FDR) of 10.24 %, a frame-to-frame distance error (DE) of 7.34 mm, a symmetric Hausdorff distance error (HD) of 10.81 mm, and a mean angular error (MEA) of 2.05°, outperforming state-of-the-art methods by 13.24 %, 15.11 %, 3.57 %, and 6.32 %, respectively.

CONCLUSION: By integrating optical flow features and deeply exploring contextual spatiotemporal dependencies, the proposed network can directly predict the relative motions between multiple frames of ultrasound images without the need for tracking, surpassing the accuracy of existing methods.

PMID:40367539 | DOI:10.1016/j.cmpb.2025.108843

J Med Chem. 2025 May 14. doi: 10.1021/acs.jmedchem.5c00536. Online ahead of print.

ABSTRACT

Explainable machine learning that identifies molecular "hot spots" for chameleonicity can guide rapid chemical design for oral absorption of beyond-rule-of-five (bRo5) drugs. Traditional in silico methods rely on computationally intensive 3D physics-based modeling or classical descriptors that do not fully explain bRo5 drug behavior. To address this, we introduced the EPSA-to-TPSA ratio (ETR) as a high-throughput measure of polarity reduction, generating data for thousands of macrocycles, PROTACs, and other bRo5s. Using this data set, we developed an explainable deep learning model to predict EPSA and locate polarity-reducing "hot spots" that influence chameleonicity. This first-of-its-kind interpretable model in the bRo5 3D domain guides chemical modifications before synthesis, helping chemists optimize physicochemical properties and design complex bRo5 drugs with improved oral bioavailability. Model insights validated by molecular dynamics enable robust, high-throughput predictions of bRo5 chameleonic behavior, building on Lipinski descriptors to establish new frameworks for complex drug design.

PMID:40367343 | DOI:10.1021/acs.jmedchem.5c00536

J Chem Theory Comput. 2025 May 14. doi: 10.1021/acs.jctc.5c00175. Online ahead of print.

ABSTRACT

Accurate modeling of protein-protein complex structures is essential for understanding biological mechanisms. Hydrogen-deuterium exchange (HDX) experiments provide valuable insights into binding interfaces. Incorporating HDX data into protein complex modeling workflows offers a promising approach to improve prediction accuracy. Here, we developed HDXRank, a graph neural network (GNN)-based framework for candidate structure ranking utilizing alignment with HDX experimental data. Trained on a newly curated HDX data set, HDXRank captures nuanced local structural features critical for accurate HDX profile prediction. This versatile framework can be integrated with a variety of protein complex modeling tools, transforming the HDX profile alignment into a model quality metric. HDXRank demonstrates effectiveness at ranking models generated by rigid docking or AlphaFold, successfully prioritizing functionally relevant models and improving prediction quality across all tested protein targets. These findings underscore HDXRank's potential to become a pivotal tool for understanding molecular recognition in complex biological systems.

PMID:40367339 | DOI:10.1021/acs.jctc.5c00175

PLoS One. 2025 May 14;20(5):e0323314. doi: 10.1371/journal.pone.0323314. eCollection 2025.

ABSTRACT

Human action recognition forms an important part of several aerial security and surveillance applications. Indeed, numerous efforts have been made to solve the problem in an effective and efficient manner. Existing methods, however, are generally aimed to recognize either solo actions or interactions, thus restricting their use to specific scenarios. Additionally, the need remains to devise lightweight and computationally efficient models to make them deployable in real-world applications. To this end, this paper presents a generic lightweight and computationally efficient Transformer network-based model, referred to as InterAcT, that relies on extracted bodily keypoints using YOLO v8 to recognize human solo actions as well as interactions in aerial videos. It features a lightweight architecture with 0.0709M parameters and 0.0389G flops, distinguishing it from the AcT models. An extensive performance evaluation has been performed on two publicly available aerial datasets: Drone Action and UT-Interaction, comprising a total of 18 classes including both solo actions and interactions. The model is optimized and trained on 80% train set, 10% validation set and its performance is evaluated on 10% test set achieving highly encouraging performance on multiple benchmarks, outperforming several state-of-the-art methods. Our model, with an accuracy of 0.9923 outperforms the AcT models (micro: 0.9353, small: 0.9893, base: 0.9907, and large: 0.9558), 2P-GCN (0.9337), LSTM (0.9774), 3D-ResNet (0.9921), and 3D CNN (0.9920). It has the strength to recognize a large number of solo actions and two-person interaction classes both in aerial videos and footage from ground-level cameras (grayscale and RGB).

PMID:40367248 | DOI:10.1371/journal.pone.0323314

PLoS One. 2025 May 14;20(5):e0312208. doi: 10.1371/journal.pone.0312208. eCollection 2025.

ABSTRACT

BACKGROUND: Gynecological cancers are among the most prevalent cancers in women worldwide. Brachytherapy, often used as a boost to external beam radiotherapy, is integral to treatment. Advances in computation, algorithms, and data availability have popularized the use of machine learning to predict patient outcomes. Recent studies have applied models such as logistic regression, support vector machines, and deep learning networks to predict specific toxicities in patients who have undergone brachytherapy.

OBJECTIVE: To develop and compare machine learning models for predicting grade 3 or higher toxicities in gynecological cancer patients treated with high dose rate (HDR) brachytherapy, aiming to contribute to personalized radiation treatments.

METHODS: A retrospective analysis was performed on gynecological cancer patients who underwent HDR brachytherapy with Syed-Neblett or Tandem and Ovoid applicators from 2009 to 2023. After applying exclusion criteria, 233 patients were included in the analysis. Dosimetric variables for the high-risk clinical target volume (HR-CTV) and organs at risk, along with tumor, patient, and toxicity data, were collected and compared between groups with and without grade 3 or higher toxicities using statistical tests. Seven supervised classification machine learning models (Logistic Regression, Random Forest, K-Nearest Neighbors, Support Vector Machines, Gaussian Naive Bayes, Multi-Layer Perceptron Neural Networks, and XGBoost) were constructed and evaluated. The training process involved sequential feature selection (SFS) when appropriate, followed by hyperparameter tuning. Final model performance was characterized using a 25% withheld test dataset.

RESULTS: The top three ranking models were Support Vector Machines, Random Forest, and Logistic Regression, with F1 testing scores of 0.63, 0.57, and 0.52; normMCC testing scores of 0.75, 0.77, and 0.71; and accuracy testing scores of 0.80, 0.85, and 0.81, respectively. The SFS algorithm selected 10 features for the highest-ranking model. In traditional statistical analysis, HR-CTV volume, Charlson Comorbidity Index, Length of Follow-Up, and D2cc - Rectum differed significantly between groups with and without grade 3 or higher toxicities.

CONCLUSIONS: Machine learning models were developed to predict grade 3 or higher toxicities, achieving satisfactory performance. Machine learning presents a novel solution to creating multivariable models for personalized radiation therapy.

PMID:40367095 | DOI:10.1371/journal.pone.0312208

Gut Microbes. 2025 Dec;17(1):2505115. doi: 10.1080/19490976.2025.2505115. Epub 2025 May 14.

ABSTRACT

Motility is a fundamental characteristic of bacteria. Distinguishing between swarming and swimming, the two principal forms of bacterial movement, holds significant conceptual and clinical relevance. Conventionally, the detection of bacterial swarming involves inoculating samples on an agar surface and observing colony expansion, which is qualitative, time-intensive, and requires additional testing to rule out other motility forms. A recent methodology that differentiates swarming and swimming motility in bacteria using circular confinement offers a rapid approach to detecting swarming. However, it still heavily depends on the observer's expertise, making the process labor-intensive, costly, slow, and susceptible to inevitable human bias. To address these limitations, we developed a deep learning-based swarming classifier that rapidly and autonomously predicts swarming probability using a single blurry image. Compared with traditional video-based, manually processed approaches, our method is particularly suited for high-throughput environments and provides objective, quantitative assessments of swarming probability. The swarming classifier demonstrated in our work was trained on Enterobacter sp. SM3 and showed good performance when blindly tested on new swarming (positive) and swimming (negative) test images of SM3, achieving a sensitivity of 97.44% and a specificity of 100%. Furthermore, this classifier demonstrated robust external generalization capabilities when applied to unseen bacterial species, such as Serratia marcescens DB10 and Citrobacter koseri H6. This competitive performance indicates the potential to adapt our approach for diagnostic applications through portable devices, which would facilitate rapid, objective, on-site screening for bacterial swarming motility, potentially enhancing the early detection and treatment assessment of various diseases, including inflammatory bowel diseases (IBD) and urinary tract infections (UTI).

PMID:40366861 | DOI:10.1080/19490976.2025.2505115

Brief Bioinform. 2025 May 1;26(3):bbaf209. doi: 10.1093/bib/bbaf209.

ABSTRACT

Accurate prediction of patient survival rates in cancer treatment is essential for effective therapeutic planning. Unfortunately, current models often underutilize the extensive multimodal data available, affecting confidence in predictions. This study presents MMSurv, an interpretable multimodal deep learning model to predict survival in different types of cancer. MMSurv integrates clinical information, sequencing data, and hematoxylin and eosin-stained whole-slide images (WSIs) to forecast patient survival. Specifically, we segment tumor regions from WSIs into image tiles and employ neural networks to encode each tile into one-dimensional feature vectors. We then optimize clinical features by applying word embedding techniques, inspired by natural language processing, to the clinical data. To better utilize the complementarity of multimodal data, this study proposes a novel fusion method, multimodal fusion method based on compact bilinear pooling and transformer, which integrates bilinear pooling with Transformer architecture. The fused features are then processed through a dual-layer multi-instance learning model to remove prognosis-irrelevant image patches and predict each patient's survival risk. Furthermore, we employ cell segmentation to investigate the cellular composition within the tiles that received high attention from the model, thereby enhancing its interpretive capacity. We evaluate our approach on six cancer types from The Cancer Genome Atlas. The results demonstrate that utilizing multimodal data leads to higher predictive accuracy compared to using single-modal image data, with an average C-index increase from 0.6750 to 0.7283. Additionally, we compare our proposed baseline model with state-of-the-art methods using the C-index and five-fold cross-validation approach, revealing a significant average improvement of nearly 10% in our model's performance.

PMID:40366860 | DOI:10.1093/bib/bbaf209

Dentomaxillofac Radiol. 2025 May 14:twaf038. doi: 10.1093/dmfr/twaf038. Online ahead of print.

ABSTRACT

OBJECTIVES: To investigate the clinical efficiency of an AI-based metal artefact correction algorithm (AI-MAC), for reducing dental metal artefacts in head and neck CT, compared to conventional interpolation-based MAC.

METHODS: We retrospectively collected 41 patients with non-removal dental hardware who underwent non-contrast head and neck CT prior to radiotherapy. All images were reconstructed with standard reconstruction algorithm (SRA), and were additionally processed with both conventional MAC and AI-MAC. The image quality of SRA, MAC and AI-MAC were compared by qualitative scoring on a 5-point scale, with scores ≥ 3 considered interpretable. This was followed by a quantitative evaluation, including signal-to-noise ratio (SNR) and artefact index (Idxartefact). Organ contouring accuracy was quantified via calculating the dice similarity coefficient (DSC) and hausdorff distance (HD) for oral cavity and teeth, using the clinically accepted contouring as reference. Moreover, the treatment planning dose distribution for oral cavity was assessed.

RESULTS: AI-MAC yielded superior qualitative image quality as well as quantitative metrics, including SNR and Idxartefact, to SRA and MAC. The image interpretability significantly improved from 41.46% for SRA and 56.10% for MAC to 92.68% for AI-MAC (p < 0.05). Compared to SRA and MAC, the best DSC and HD for both oral cavity and teeth were obtained on AI-MAC (all p < 0.05). No significant differences for dose distribution were found among the three image sets.

CONCLUSION: AI-MAC outperforms conventional MAC in metal artefact reduction, achieving superior image quality with high image interpretability for patients with dental hardware undergoing head and neck CT. Furthermore, the use of AI-MAC improves the accuracy of organ contouring while providing consistent dose calculation against metal artefacts in radiotherapy.

ADVANCES IN KNOWLEDGE: AI-MAC is a novel deep learning-based technique for reducing metal artefacts on CT. This in-vivo study first demonstrated its capability of reducing metal artefacts while preserving organ visualization, as compared with conventional MAC.

PMID:40366748 | DOI:10.1093/dmfr/twaf038

Bioinformatics. 2025 May 14:btaf301. doi: 10.1093/bioinformatics/btaf301. Online ahead of print.

ABSTRACT

MOTIVATION: Crohn's disease (CD) exhibits substantial variability in response to biological therapies such as ustekinumab (UST), a monoclonal antibody targeting interleukin-12/23. However, predicting individual treatment responses remains difficult due to the lack of reliable histopathological biomarkers and the morphological complexity of tissue. While recent deep learning methods have leveraged whole-slide images (WSIs), most lack effective mechanisms for selecting relevant regions and integrating patch-level evidence into robust patient-level predictions. Therefore, A framework that captures local histological cues and global tissue context is needed to improve prediction performance.Ustekinumab (UST) is a relatively recent biologic agent used in the treatment of Crohn's Disease (CD). Clinical studies on the treatment response of UST are relatively scarce. However, its efficacy varies among CD patients, highlighting the need for accurate to prediction of its treatment response. In this paper, We developed an artificial intelligence (AI) model based on whole-slide images (WSIs) and weakly supervised learning to predict the treatment response of UST in CD patients.

RESULTS: We propose a novel clustering-enhanced weakly supervised learning framework to predict UST treatment response from pre-treatment WSIs of CD patients. First, patches from WSIs were encoded using a pre-trained vision foundation model, and k-means clustering was applied to identify representative morphological patterns. Discriminative patches associated with treatment outcomes were selected via a DenseNet-based classifier, with Grad-CAM used to enhance interpretability. To aggregate patch-level predictions, we adopted a multi-instance learning approach, from which whole-slide features were extracted using both patch likelihood histograms and bag-of-words representations. These features were subsequently used to train a classifier for final response prediction. Experimental results on an independent test set demonstrated that our WSI-level model achieved superior predictive performance with an AUC of 0.938 (95% CI: 0.879-0.996), sensitivity of 0.951, and specificity of 0.825, outperforming baseline patch-level models. These findings suggest that our method enables accurate, interpretable, and scalable prediction of biological therapy response in CD, potentially supporting personalized treatment strategies in clinical settings.402 tissue samples from CD patients treated with UST were categorized into non-response and response groups based on clinical outcomes. Initially, we selected relevant patches from WSIs, then patch-level treatment efficacy predictions were constructed using deep learning methods. Subsequently, pathological features generated by patches predict results aggregation were combined with various machine learning algorithms to develop a WSI-level AI model. This enables automatic prediction of UST treatment response for CD from label-free WSIs. Our model demonstrated competitive performance in predicting UST treatment response, with AUC of 0.866 (95%CI:0.865-0.867), sensitivity of 0.807, and specificity of 0.746 at the patch-level in the independent testset. The multi-instance learning (MIL) method, which aggregates patch-level result features to predict WSI-level treatment response, further enhanced the model's performance. Our model achieved an AUC of 0.938 (95%CI:0.879-0.996), with a sensitivity of 0.951 and a specificity of 0.825 in the independent test set, surpassing patch-level prediction performance.The AI model developed in this study, based on pre-treatment biopsy pathology images, accurately predicts UST treatment response in CD patients and can potentially be extended to other similar prediction tasks.

AVAILABILITY AND IMPLEMENTATION: https://github.com/caicai2526/USTAIM.

SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

PMID:40366737 | DOI:10.1093/bioinformatics/btaf301

Discov Oncol. 2025 May 14;16(1):763. doi: 10.1007/s12672-025-02590-4.

ABSTRACT

OBJECTIVE: To assess the publications' bibliographic features and look into how the advancement of artificial intelligence (AI) and its subfields in radiomics has affected the growth of oncology.

METHODS: The researchers conducted a search in the Web of Science (WoS) for scientific publications in cancer pertaining to AI and radiomics, published in English from 1 January 2015 to 31 December 2024.The research included a scientometric methodology and comprehensive data analysis utilising scientific visualization tools, including the Bibliometrix R software package, VOSviewer, and CiteSpace. Bibliometric techniques utilised were co-authorship, co-citation, co-occurrence, citation burst, and performance Analysis.

RESULTS: The final study encompassed 4,127 publications authored by 5,026 individuals and published across 597 journals. China (2087;50.57%) and USA (850;20.6%) were the two most productive countries. The authors with the highest publication counts were Tian Jie (60) and Cuocolo Renato (30). Fudan University (169;4.09%) and Sun Yat-sen University (162;3.93%) were the most active institutions. The foremost journals were Frontiers in Oncology and Cancer. The predominant author keywords were radiomics, artificial intelligence, and oncology research.

CONCLUSION: Investigations into the integration of AI with radiomics in oncology remain nascent, with numerous studies concentrating on biology, diagnosis, treatment, and cancer risk evaluation.

PMID:40366503 | DOI:10.1007/s12672-025-02590-4

Nucleic Acids Res. 2025 May 14:gkaf416. doi: 10.1093/nar/gkaf416. Online ahead of print.

ABSTRACT

Atomic structure modeling is a crucial step in determining the structures of protein complexes using cryo-electron microscopy (cryo-EM). This work introduces DEMO-EMol, an improved server that integrates deep learning-based map segmentation and chain fitting to accurately assemble protein-nucleic acid (NA) complex structures from cryo-EM density maps. Starting from a density map and independently modeled chain structures, DEMO-EMol first segments protein and NA regions from the density map using deep learning. The overall complex is then assembled by fitting protein and NA chain models into their respective segmented maps, followed by domain-level fitting and optimization for protein chains. The output of DEMO-EMol includes the final assembled complex model along with overall and residue-level quality assessments. DEMO-EMol was evaluated on a comprehensive benchmark set of cryo-EM maps with resolutions ranging from 1.96 to 12.77 Å, and the results demonstrated its superior performance over the state-of-the-art methods for both protein-NA and protein-protein complex modeling. The DEMO-EMol web server is freely accessible at https://zhanggroup.org/DEMO-EMol/.

PMID:40366028 | DOI:10.1093/nar/gkaf416