• We introduce PGSFNet for real-world scene text image super-resolution. • Adaptive Frequency Modulator is proposed to extract informative frequency components. • Text Information Enhancement module is designed to incorporate text priors. • We develop a Sobel loss to guide optimization towards sharper text details. • PGSFNet is shown to achieve superior performance on public text image datasets. Real-world scene text image super-resolution is challenging due to complex writing strokes, random text distribution, and diverse scene degradations. Existing text super-resolution methods focus on pure text images or fixed-size single-line text, which limits their practical utility. To address that, we propose a Prompt-Guided Selective Frequency super-resolution Network (PGSFNet). Our unique bicephalous neural model comprises a super-resolution branch and a prompt guidance branch. The latter specifically helps in leveraging text content-aware information priors. To that end, we propose a Text Information Enhancement module. To exploit selective frequency information present in the image, PGSFNet employs a proposed Adaptive Frequency Modulator fused with multi-attention structures. Considering the criticality of text edges in our task, we also propose a tailored text edge perception loss. Extensive experiments on the standard open real-world scene text image datasets demonstrate remarkable performance of our method, achieving up to 8.75% PNSR gain for × 2 and 2.28% SSIM gain for × 4 super-resolution on the Real-CE dataset. Our code will be made public at https://github.com/holastq/PGSFNet .

Graph adiabatic diffusion neural networks for distribution-shift breast tumor image classification.

Bronchoscopy is routinely conducted in pulmonary clinics and intensive care units, but navigating the complex branching of the respiratory tract remains challenging. This paper introduces BronchoLumen, a real-time YOLO-based system for detecting bronchial orifices in video bronchoscopy, aiming to assist navigation and CAD systems. The paper investigates if bronchial orifices can be robustly detected across image domains using state-of-the-art object detection and a limited set of public image data. The study includes the description and comparison of YOLOv8, a widely adopted architecture, and YOLOv12, a more recent architecture integrating attention-based modules to improve spatial reasoning. Both models are trained and tested solely on publicly available datasets comprising different image domains. A comparison of both models is conducted based on the common metrics mAP@0.5 and mAP@0.5:0.9 with the latter emphasizing localization accuracy. For YOLOv8 we obtained a mAP@0.5 of 0.91 on an in-domain and 0.68 on a cross-domain test set. YOLOv12 achieved 0.84 and 0.68 respectively with slightly better localization accuracy with mAP@0.5:0.9 of 0.48 and 0.26 compared to YOLOv8 with 0.45 and 0.25. Challenges like motion blur and low contrast occasionally entailed uncertainties but the system demonstrated overall robustness in most scenarios. BronchoLumen is an open-weight, YOLO-based solution for bronchial orifice detection offering high accuracy and efficiency across multiple image domains. While the more recent YOLOv12 achieves better localization accuracy, we observed a slightly worse precision. The models have been made publicly available to foster further research in bronchoscopy navigation.

Deformable image registration plays an essential role in various medical image tasks. Existing deep learning-based deformable registration frameworks primarily utilize convolutional neural networks (CNNs) or Transformers to learn features to predict the deformations. However, the lack of semantic information in the learned features limits the registration performance. Furthermore, the similarity metric of the loss function is often evaluated only in the pixel space, which ignores the matching of high-level anatomical features and can lead to deformation folding. To address these issues, in this work, we proposed LDM-Morph, an unsupervised deformable registration algorithm for medical image registration. LDM-Morph integrated features extracted from the latent diffusion model (LDM) to enrich the semantic information. Additionally, a latent and global feature-based cross-attention module (LGCA) was designed to enhance the interaction of semantic information from LDM and global information from multi-head self-attention operations. Finally, a hierarchical metric was proposed to evaluate the similarity of image pairs in both the original pixel space and latent-feature space, enhancing topology preservation while improving registration accuracy. Extensive experiments on four public 2D cardiac image datasets, two 3D image datasets, show that the proposed LDM-Morph framework outperformed existing state-of-the-art CNNs-and Transformers-based registration methods regarding accuracy with comparable topology preservation and computational efficiency. Our code is publicly available at: https://github.com/wujiong-hub/LDM-Morph.

Ultrasound image segmentation remains a significant challenge due to inherent low contrast and blurred anatomical boundaries. Fully supervised deep learning approaches require extensive annotated datasets, which are costly and labor-intensive to acquire. This study aims to develop an effective semi-supervised segmentation framework for ultrasound images with limited annotations. We propose a novel semi-supervised segmentation framework tailored for ultrasound images, leveraging frequency component augmentation and edge mask enhancement to promote structural consistency between weakly and strongly augmented inputs. Specifically, discrete wavelet transform (DWT) is used to decompose ultrasound images into low-frequency and high-frequency sub-bands. A high-frequency component replacement strategy is introduced for strongly augmented images, and an edge mask enhancement module is designed to further emphasize anatomical boundaries. Experiments conducted on 3 public fetal ultrasound imaging segmentation datasets-PSFHS, HC18, and CCAUI-demonstrate that our method achieves average Dice similarity coefficients (DSC) of 0.81 and 0.91, respectively, using only 10 annotated images. This represents a 2-3% DSC improvement over existing semi-supervised methods such as FixMatch. Ablation studies confirm the effectiveness of both the high-frequency augmentation and edge enhancement components. The proposed framework offers a promising direction for ultrasound image segmentation in settings with limited annotations, effectively improving segmentation accuracy by combining frequency-domain augmentation and edge-aware enhancement. Code will be available at https://github.com/apple1986/WTEM-SemiSeg.

This study presents a novel approach using graph neural networks to predict the risk of internal bleeding using vessel maps derived from patient CT and MRI scans, aimed at enhancing the realism of surgical simulators for emergency scenarios such as trauma, where rapid detection of internal bleeding can be lifesaving. First, medical images are segmented and converted into graph representations of the vasculature, where nodes represent vessel branching points with spatial coordinates and edges encode vessel features such as length and radius. Due to no existing dataset directly labeling bleeding risks, we calculate the bleeding probability for each vessel node using a physics-based heuristic, peripheral vascular resistance via the Hagen-Poiseuille equation. A graph attention network is then trained to regress these probabilities, effectively learning to predict hemorrhage risk from the graph-structured imaging data. The model is trained using a tenfold cross-validation on a combined dataset of 1708 vessel graphs extracted from four public image datasets (MSD, KiTS, AbdomenCT, CT-ORG) with optimization via the Adam optimizer, mean squared error loss, early stopping, and L2 regularization. Our model achieves a mean R-squared of 0.86, reaching up to 0.9188 in optimal configurations and low mean training and validation losses of 0.0069 and 0.0074, respectively, in predicting bleeding risk, with higher performance on well-connected vascular graphs. Finally, we integrate the trained model into an immersive virtual reality environment to simulate intra-abdominal bleeding scenarios for immersive surgical training. The model demonstrates robust predictive performance despite the inherent sparsity of real-life datasets.

This paper presents a new mathematical model for designing a reverse logistics network (RLN) that considers carbon emissions, energy recovery, and shipping frequency. This model minimizes operating costs and environmental impacts through a multi-objective mixed-integer programming (MOMIP). Given the significance of reverse logistics (RL) in waste management, this mathematical model aims to reduce logistics system costs and lower emissions of environmental pollutants. Within the RL system, the processes of collection, reconstruction, recycling, energy recovery, and disposal of products operate as an independent network. This system analyzes the physical flow of goods from consumers to primary suppliers, encompassing the processing, control, conversion, and maintenance of the flow of raw materials, parts, finished products, inventory, and capital. To solve this model, mathematical methods utilize the weighted sum method (WSM). The results of solving the mathematical model reveal that altering the weighting of the objectives produces a decreasing trend for the first objective function and an increasing trend for the second objective function, thus confirming the trade-off between the model’s objectives. In companies, this results in a more efficient use of resources and improved services for customers. Moreover, emphasizing economic and environmental responsibility enhances the public image of companies. The novelty of this work lies in its comprehensive and integrated approach to environmentally friendly RL, bridging the gaps between cost optimization, carbon footprint reduction, and recycling efficiency. Its contributions are both theoretical and practical.

Secure data transmission within digital images remains a critical challenge due to vulnerabilities to key interception and steganalysis attacks. Traditional steganographic schemes often require shared keys, pre-trained models, or prior coordination, which limits their practical deployment in open environments without prior synchronization or shared secrets. This paper introduces a symmetric dual-key encryption-steganography hybrid, inspired by one-time pad (OTP) principles, that enables secure image-based communication without any key exchange or prior shared knowledge. The method achieves high secrecy and imperceptibility, embedding hidden data without introducing visible distortions or statistical artifacts. The approach is lightweight, general, and does not depend on training or image-specific assumptions. Experimental validation on 100 natural images demonstrates strong resilience to advanced steganalysis, high visual quality (SSIM > 0.97, PSNR > 40 dB), and secure hidden data transmission. These results highlight the method's practical value as a robust and transparent solution for sensitive image-based communication, without the limitations of prior coordination or machine learning infrastructure.

Few-shot medical image classification is a highly challenging problem in computer-aided diagnosis, with the central difficulty being enabling deep models to learn discriminative features conducive to classification from limited labeled samples. Vision transformers (ViTs) have recently demonstrated outstanding performance across various visual tasks. However, owing to their large parameter counts and dependence on massive pretraining data, ViTs are prone to overfitting in sample-scarce scenarios typical of few-shot learning. Parameter-efficient fine-tuning (PEFT) techniques, such as low-rank adaptation (LoRA), have alleviated some of these issues. However, conventional PEFT approaches still encounter difficulties in complex medical image classification tasks. To address this, this study proposes a general fine-tuning framework called a hierarchical probing and fusion network (HPF-Net), which integrates three core innovations to allow smarter and more efficient adaptation for few-shot medical image classification. First, a Fisher information-driven layer selection strategy strengthens the layer-selection robustness in few-shot settings. Subsequently, the attention-guided multiscale fusion module aligns and improves the features drawn from the selected critical layers. Subsequently, LoRA is incorporated into this efficient fine-tuning pipeline to reduce the parameter overhead while improving the accuracy. Extensive experiments on the public few-shot medical image benchmark, the medical imaging meta-dataset, demonstrated that HPF-Net significantly outperformed baseline methods, and ablation studies validated the necessity of each proposed component. The source code will be released upon acceptance.

Evaluating consumer expectations and perceptions of established services is crucial to assessing the quality of drive through pharmacy services. Therefore, this study aimed to assess public perceptions of pharmacist image and the perceived advantages and disadvantages in response to the establishment of drive through pharmacy services. A cross-sectional quantitative study was conducted among individuals living in Riyadh, the capital of Saudi Arabia, in 2024. The study took place from March to July 2024. The questionnaire consisted of 5 sections and a total of 28 items. The mean age of the respondents was 58.4 years (standard deviation = 10.4; range: 20–83 years). Approximately 60% of respondents believed that pharmacists maintain an appropriate balance between patient care and the business aspects of their profession. Moreover, 55.2% agreed that prescriptions are likely to be processed more quickly at drive through pharmacies compared with traditional settings. Notably, 88.4% (n = 296) recognized drive through pharmacies as particularly beneficial for serving individuals who are ill, elderly, or physically disabled. The mean score for the public image of pharmacists was 10.9 ± 2.1 (median = 11), while the mean perceived benefit score for drive through pharmacies was 26.5 ± 4.06 (median = 28). In contrast, the mean score for perceived drawbacks was 24.1 ± 5.4 (median = 24). Overall, the current study provided a preliminary insight into people’s perceptions of the role of pharmacists, as well as the benefits and drawbacks of drive through pharmacy services, which had not yet been published. A majority of respondents felt that pharmacists strike a good balance and prioritize patient health over commercial matters.

The article examines the digital political communication of a new generation of right-wing populist leaders in the European Union. The aim of the study is to identify the mechanisms through which systematic digital communication transforms into a stable digital reputation and becomes a significant resource for political mobilisation. The working hypothesis suggests that regular and personalised communication by right-wing populist leaders on digital platforms contributes to strengthening their public image, shaping a resilient digital reputation, and increasing their appeal among younger audiences. The study draws on contemporary approaches to populism as a style of political mobilisation, as well as on concepts of digital political communication, political personalisation, and the mediatisation of the public sphere. Digital reputation is conceptualised as the outcome of discursive, visual, and communicative practices formed in the online environment under the influence of the algorithmic logic of platforms. The methodological framework is based on a qualitative approach, including case studies, comparative analysis, and elements of digital discourse analysis. The empirical basis consists of three cases of young right-wing populist leaders: Jordan Bardella (France), Tom Van Grieken (Belgium), and Rita Matias (Portugal). The analysis focuses on the choice of digital platforms, communication styles and formats, dominant narratives, modes of leader representation, and mobilisation effects. The findings demonstrate that social media function as the central channel of political communication for right-wing populists, enabling a high degree of personalisation, emotionalisation, and direct interaction with audiences. Despite differences in national contexts, digital strategies exhibit similar discursive elements, most notably the reproduction of the populist dichotomy between “the people” and “the elite”. The identified differences point to the variability of models of digital reputation formation, ranging from personalised leadership to symbolic reputational positioning. The article concludes that digital reputation constitutes an important factor in political mobilisation and in the transformation of party leadership under conditions of digitalisation.

Background: The global shortage of healthcare workers continues to grow each year. In particular, low nurse staffing levels are known to be associated with adverse patient outcomes. Helping adolescents understand the nursing profession beyond stereotypical societal perceptions—and recognize its full range of roles—may make nursing a more attractive career option. However, little is known about the perceptions of adolescents who are in the process of making career choices. Objective: This study aims to determine the perceptions of high school adolescents in Türkiye regarding the image of the nursing profession. Method: The sample of this cross-sectional study consisted of 581 high school adolescents in Türkiye. Data were collected using the Descriptive Characteristics Form and the Adolescents’ Perceptions of the Image of Nursing Scale (APNIS). In addition to descriptive statistics, independent-samples t-tests, one-way ANOVA, and Chi-Square analyses were performed. Results: The findings indicated that adolescents generally hold a positive perception of the nursing profession, with higher scores in the professional image, perception, and care and therapeutic role subscales, and lower scores in the communication, informative role, and healing environment subscales. A total of 65.7% of the adolescents reported that they did not intend to choose nursing as a career. The intention to choose nursing as a career was higher among adolescents who did not have a nurse in their family (p < 0.05). Conclusions: The results indicate that positive perceptions of the nursing image alone may not be sufficient in adolescents’ career decision-making and underscore the importance of presenting the profession’s roles in a comprehensive and realistic manner. While adolescents with a nurse in the family demonstrated more positive perceptions of the nursing profession (higher APNIS scores), their intention to choose nursing as a career was lower compared to those without a nurse in the family. These findings suggest that familiarity with the profession may positively influence professional image while simultaneously reducing career intention, possibly due to greater awareness of occupational challenges. The findings provide important insights into how adolescents form their perceptions of the image of nursing and may contribute to future research and educational initiatives aimed at increasing interest in the nursing profession.

LRF-CNN: An explainable lightweight receptive field-based CNN for colorectal cancer histopathological image classification

Functional connectivity analysis of brain networks has provided valuable insights for brain disorders diagnosis. Recent studies have focused on collaborative learning with multiple brain atlases to overcome the limitations of single-atlas information. However, these approaches often overlook sufficient interaction and consistency among multiple atlases, as well as information redundancy resulting from multi-atlas fusion. We propose a unified multi-atlas learning framework (BrainUMA) with hyper-connectivity network learning for brain disorders diagnosis, which consists of two key stages: hyper-connectivity network construction, and cross-atlas HCN interactions. We employ FCN for hyper-connectivity network construction and propose a novel hyper-connectivity network construction strategy, which includes both the hypergraph structure construction and node feature learning. Meanwhile, to sufficiently model interactions across multiple atlases, we propose a feature disentanglement method that disentangles disease-related information with hyperedge-aware hypergraph convolutional networks. We introduce two loss functions: an atlas-based contrastive loss and a class-consistency loss to guide the disentanglement processes. We evaluate our model on the public Autism Brain Imaging Data Exchange (ABIDE) dataset to demonstrate the effectiveness of the proposed model and investigate the optimal combination of brain atlases. Our results shed new light on the importance of exploiting the relationship among by disentanglement for improving multi-atlas disease diagnosis. In addition, our model provides deeper insights into disease interpretability, including atlas properties and critical brain regions. Our code is publicly available at https://github.com/MortonHao/BrainUMA .

Titanium three-dimensional (3D) printed triflange cups are an established treatment option for severe pelvic discontinuity, but are costly and, when produced offsite, involve weeks of surgical delay. Three-dimensional printing of polyetheretherketone (PEEK) at the point of care (POC) has recently been 510(k) cleared by the Food and Drug Administration for cranioplasty to reduce implant cost and time to treatment. We asked in this pilot study whether 3D-printed triflange cups from high-strength polyaryletherketone (PAEK) polymers like PEEK could be produced with sufficient strength for revision total hip arthroplasty and whether annealing could improve cup strength. For proof-of-concept, two representative triflange cup designs were developed based on a set of 49 full-body computed tomography scans from the public New Mexico Decedent Image Database repository. The triflange cups (n = 3 per design) were additively manufactured using an industrial 3D printer (F421) and AM 200 PAEK filament. A second set of cups (n = 3 per design) was annealed in an effort to improve strength. A customizable fixture was developed for mechanical testing of the personalized cups, which were loaded to failure in an MTS load frame. As-printed (nonannealed) polymer cups failed due to deformation followed by layer separation, whereas the annealed cups fractured. The ultimate load (mean ± SD) for the two nonannealed triflange cup designs was 10.2 ± 0.02 kN and 11.0 ± 1.9 kN, respectively. After annealing, the strength of the two designs was 2.60 ± 0.44 kN and 10.5 ± 2.7 kN. The nonannealed PAEK cups exceeded the strength published for a titanium 3D-printed design (5.4 kN). The promising results of this pilot study demonstrate proof-of-concept and suggest that PAEK polymers such as PEEK may be suitable biomaterials for future triflange cup applications. Additional research is needed to incorporate porosity into the printed designs and further refine the mechanical test model.

ABSTRACT This study examines how international political developments during the long 1990s (1989–2001) influenced the Poetry International Festival, focusing on both its programming and public image. It asks whether changing geopolitical conditions affected the selection of poets and the themes addressed in festival discussions. To analyse the festival’s image, the study employs Jérôme Meizoz’s concept of posture, adapting it so it encompasses a set of auto- and hetero-representations of a collective actor. The analysis is based on five festival editions (1989, 1990, 1994, 2001, and 2002). The findings show that changes in the festival’s auto- and hetero-representations were complex and uneven. In the early 1990s, Poetry International presented itself as a strongly politicized and activist festival, offering a ‘safe haven’ for dissident poets in a divided world. The fall of the Berlin Wall did not prevent Western intellectuals from reproducing Cold War – style divisions between Western Europe and other regions, particularly Central and Eastern Europe, even though non-Western voices sometimes challenged this view. In the early 2000s, the festival gradually shifted towards a more globalized and less explicitly politicized profile, reflecting broader post – Cold War trends in world poetry.

This paper argues that musical reception is not merely an aesthetic experience but a sociomusical process shaped by the artist’s identity and political stance within hegemonic relations. For Frith (1996), listeners’ emotional alliances formed with the performer and the performer’s other listeners are crucial for understanding how the bond with the artist is constructed. In conditions of heightened political intensity, the artist’s political stance, articulated through their practices and statements as part of their public image, directly influences musical reception. Following Shiner’s (2001) account of the “invention of art,” it demonstrates that the concepts of art and the artist are historically linked to the rise of the bourgeoisie, highlighting how the connotations of “genius” and “freedom” have served as ideological foundations defining the artist’s social role. In this regard, the paper examines anecdotes about the image of the artist, which Kris and Kurz (1934/1979) trace back to antiquity. By focusing on “listening to the artist’s identity,” this paper demonstrates how the artist’s political position transforms musical reception itself. The artist emerges as a symbolic center of power, shaped through discursive processes that extend beyond the individual subject, carrying historical and social meanings and representing political positions. Within the framework of Gramsci’s (1947/2021) theory of hegemony, it is argued that artists have the potential to produce either “consent” or “resistance” within the hegemonic order, as organic or traditional intellectuals. This paper presents a theoretical/critical review based on an interpretive reading of key literature and publicly available materials. Finally, it concludes that, because of the historical and ideological construction of the concepts of art and the artist, the artist’s political stance can influence the listener’s aesthetic experience. The paper highlights that the political alliance in which the artist, as a hegemonic subject, participates structures the emotional alliances listeners form with the artist. Thus, from a Gramscian perspective, artists – who are intellectuals by nature – can become key subjects influencing musical reception by assuming the organic or traditional intellectual roles within the hegemonic order.

Camera augmentation: enabling uncalibrated stereo matching of minimally invasive surgery images by training from the wealth of public synthetic image datasets

A deep cardiac motion field analysis approach via global 2nd-order kinematic graph modeling.

Abstract Ore image segmentation plays a vital role in mineral processing, directly affecting the accuracy of crushing quality assessment and particle size analysis. However, the accuracy of traditional segmentation techniques is severely compromised by three inherent challenges in ore images: the wide size range of particles, significant color variations within individual ores, and indistinct boundaries between agglomerated particles, leading to imprecise segmentation results. To overcome these challenges, this study introduces LAES-UNet, an integrated network based on EfficientSAM, which combines a pre-trained EfficientSAM encoder with a multi-level decoder. The proposed architecture incorporates three tailored modules: the Local-Global Hierarchical Interaction (LGHI) for multi-scale feature enhancement, the Adaptive Spatial Feature Refinement (ASFR) for adaptive weighting across color-variable regions, and the Edge Focusing Module (EFM) for explicit edge and fine-detail perception. Experiments were conducted on a self-built conveyor belt ore image dataset containing 148 manually annotated images, which were cropped into 296 non-overlapping samples, as well as on a public mineral image benchmark. The results show that LAES-UNet achieves the best overall segmentation performance among the compared methods, with up to 2.8% higher IoU and consistently improved boundary delineation. Furthermore, a kernel density estimation (KDE)-based particle size distribution fitting method verifies the practical value of the segmentation results in quantitative ore particle size analysis. Overall, LAES-UNet delivers a generalizable solution for automated, high-precision particle size measurement within intelligent mineral processing systems.