Gadolinium-based contrast agents (GBCAs) are widely used in magnetic resonance imaging (MRI) to enhance lesion detection and characterisation, particularly in the field of neuro-oncology. Nevertheless, concerns regarding gadolinium retention and accumulation in brain and body tissues, most notably for diseases that require close monitoring and frequent GBCA injection, have led to the need for strategies to reduce dosage. In this study, a deep learning framework is proposed for the virtual contrast enhancement of full-dose post-contrast T1-weighted MRI images from corresponding low-dose acquisitions. The contribution of the presented model is its utilisation of longitudinal information, which is achieved by incorporating a prior full-dose MRI examination from the same patient. A comparative evaluation against a non-longitudinal single session model demonstrated that the longitudinal approach significantly improves image quality across multiple reconstruction metrics. Furthermore, experiments with varying simulated contrast doses confirmed the robustness of the proposed method. These results emphasize the potential of integrating prior imaging history into deep learning-based virtual contrast enhancement pipelines to reduce GBCA usage without compromising diagnostic utility, thus paving the way for safer, more sustainable longitudinal monitoring in clinical MRI practice.

Decision-making for maxillary sinus floor elevation (MSFE) surgery in patients with low residual bone height (<4 mm) presents significant challenges, particularly in selecting surgical approaches and predicting intraoperative mucosal perforation. Traditional methods rely heavily on physician experience, lack standardization and objectivity, and often fail to meet the demands of precision medicine. This study aims to build an intelligent decision-making system based on deep learning to optimize surgical selection and predict the risk of mucosal perforation, providing clinicians with a reliable auxiliary tool. This study retrospectively analysed the cone-beam computed tomography imaging data of 79 patients who underwent MSFE and constructed a three-dimensional (3D) deep-learning model based on the overall CT data of the patients for surgical procedure selection and prediction of mucosal perforation. The model innovatively introduced the Convolutional Block Attention Module mechanism and depthwise separable convolution technology to enhance the model's ability to capture spatial features and computational efficiency. The model was rigorously trained and validated on multiple datasets, with visualization achieved through attention heatmaps to improve interpretability. The modified EfficientNet model achieved an F1 score of 0.6 in the procedure decision task of MSFE. For predicting mucosal perforation, the improved ResNet model achieved an accuracy of 0.8485 and an F1-score of 0.7273 on the mixed dataset. In the experimental group, the improved ResNet model achieved an accuracy of 0.8235, a recall of 0.7619, and an F1-score of 0.7302. In the control group, the model also maintained stable performance, with an F1-score of 0.6483. Overall, the 3D convolutional model enhanced the accuracy and stability of mucosal perforation prediction by leveraging the spatial features of cone-beam computed tomography imaging, demonstrating a certain degree of generalization capability. This study is the first to construct a deep learning-based 3D intelligent decision-making model for MSFE. These findings confirm the model's effectiveness in surgical decision-making and in predicting the risk of mucosal perforation. The system provides an objective decision-making basis for clinicians, improves the standardization level of complex case management, and demonstrates potential for clinical application.

Accurate detection of anatomical landmarks from radiographic images is critical for total hip arthroplasty (THA) surgical planning and postoperative evaluation. However, existing methods face significant challenges in unstructured data, such as irregular patient postures or occluded landmarks, which hinder their robustness and reliability. This study aims to develop an advanced deep learning framework to address these challenges, by leveraging uncertainty estimation to handle unstructured data and assigning uncertainty scores to predicted landmarks, thereby alerting clinicians to focus on these results. We propose Unstructured X-ray - High-Resolution Net (UNSX-HRNet), a framework that integrates high-resolution networks with uncertainty estimation based on anatomical relationships to predict landmarks without relying on a fixed number of points. The method suppresses low-certainty landmarks to accurately handle unstructured data while highlighting the certainty level of each landmark to provide correction guidance. The model was trained and tested on both structured and unstructured datasets, with performance evaluated using multiple precision metrics. Experimental results demonstrate that UNSX-HRNet achieves improvement, exceeding 60 % across multiple evaluation metrics when applied to unstructured datasets. On structured datasets, the framework maintains high performance, showcasing its robustness and adaptability across varying data conditions. UNSX-HRNet offers a reliable and automated solution for THA landmark detection, addressing the challenges of unstructured data through uncertainty-aware predictions. This approach not only improves accuracy but also provides actionable insights for clinicians, contributing to the development of AI-driven expert systems for surgical planning and monitoring.

The aim of this work is the generation of realistic synthetic mammograms, using as an input of the imaging acquisition simulation process digital anthropomorphic phantoms, reconstructed from sets of dedicated breast computed tomography (BCT) images from different patients. The voxel-based structure and the segmentation into fibroglandular, adipose and skin tissues are performed through trivariate tensor-product B-spline approximation and morphological operations. The obtained phantoms can be modified by means of geometrical transformations that replicate typical breast shape deformities, and by locally introducing virtual masses and calcifications. After simulating biomechanical compression of the 3D breast phantoms, we generate the mammograms in both craniocaudal (CC) and mediolateral oblique (MLO) views, modelling the x-ray interaction with breast tissues with a Monte Carlo approach implemented in the in silico breast imaging pipeline VICTRE. The methodology proposed here can contribute to the creation of synthetic mammogram databases, to be used for in silico testing of diagnostic and therapeutic techniques, as well as for the validation of artificial intelligence (AI) systems in diagnostic imaging and cancer screening. The great advantage is that, from a single BCT scan, it is possible to generate multiple realistic mammograms, with different anatomical features, in terms of breast shape and size, and type and location of lesions.

X-rays are the most commonly used method for diagnosing Legg–Calvé–Perthes disease (LCPD) in children. However, in early-stage LCPD, X-rays only show subtle changes, making diagnosis highly dependent on the radiologist’s experience. The aim of this study was to develop and validate a combined radiomics and deep learning model based on anteroposterior and frog-leg lateral X-rays for identifying early LCPD patients. We retrospectively collected imaging data of children diagnosed with early LCPD and normal control groups from two centers between January 2013 and December 2023. Logistic regression (LR), Support vector machine (SVM), and Extreme gradient boosting (XGBoost) algorithms were used to build radiomics and Deep Learning (DL) models, and the performance of individual models was compared. The development of the ensemble model involved a stacking strategy to integrate the best radiomics and DL models. The diagnostic performance of the combined model was compared with that of radiologists of varying experience levels. Model evaluation was conducted using metrics such as the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Finally, the model was validated using calibration and clinical decision curves. A total of 200 early LCPD hips (Center A, <i>n</i> = 157; Center B, <i>n</i> = 43) and 236 normal hips (Center A, <i>n</i> = 188; Center B, <i>n</i> = 48) were included. Among the individual radiomics and DL models, the XGBoost algorithm performed the best: the radiomics model achieved an AUC of 0.845 (95% CI, 0.758–0.933), and the DL model achieved an AUC of 0.848 (95% CI, 0.766–0.929). The ensemble model’s performance was further improved, with an AUC of 0.878 (95% CI, 0.810, 0.945). The combined model significantly outperformed junior radiologists. Calibration and clinical decision curves demonstrated that the combined model had high predictive value. The integrated radiomics and DL model using both anteroposterior and frog-leg lateral X-rays demonstrated superior performance over individual radiomics or DL approaches, highlighting its potential as an effective tool for early screening of LCPD. The online version contains supplementary material available at 10.1186/s12891-025-09189-4.

This review highlights the evolving role of positron emission tomography (PET) in quantifying myocardial blood flow (MBF) and myocardial flow reserve (MFR) and its expanding clinical impact. The relative nature of perfusion assessment with single photon emission computed tomography often underestimates disease in patients with multivessel or microvascular involvement. Positron emission tomography (PET) enables absolute quantification of myocardial blood flow (MBF) and myocardial flow reserve (MFR), which provides deeper insights into coronary physiology. PET-derived MBF and MFR have shown clear diagnostic and prognostic value across a broad spectrum of conditions, including obstructive coronary artery disease, ischemia and angina without obstructive coronary artery disease, post-heart transplant cardiac allograft vasculopathy surveillance, diabetes, hypertension, and systemic inflammatory diseases. Impaired flow reserve consistently predicts adverse outcomes, even in the absence of visible perfusion defects. Newer tracers such as ¹⁸F-flurpiridaz, with their favorable kinetics and logistical advantages, are poised to expand clinical accessibility. At the same time, innovations such as artificial intelligence-driven analysis and total-body PET promise greater reproducibility and efficiency, further integrating flow assessment into everyday practice. Professional society guidelines now recommend routine incorporation of flow quantification into stress PET imaging, yet barriers remain, including limited access, heterogeneity in protocols, and a need for outcome-driven trials. As technology and evidence evolve, PET-based flow quantification is positioned to become an essential tool in precision cardiovascular care, bridging the gap between physiology and clinical decision-making.

Despite major advances in short-term outcomes after heart transplantation, long-term survival remains limited by chronic allograft dysfunction, with cardiac allograft vasculopathy (CAV) being the leading cause of late graft failure and an important cause of all-cause mortality. CAV is a unique and multifactorial form of transplant coronary vasculopathy, driven by a complex interplay of alloimmune responses, innate immune activation, and traditional cardiovascular risk factors. Recent insights from deep profiling of human allograft tissue have revealed the key roles of locally sustained T- and B-cell-mediated inflammation, macrophage-natural killer cell interactions, and chronic immune activation within the graft. These discoveries challenge prior models of systemic immune monitoring and highlight the importance of spatially organized, intragraft immune processes. In parallel, the diagnostic landscape of CAV is rapidly evolving. High-resolution imaging techniques such as optical coherence tomography, and advanced non-invasive tools including coronary computed tomography angiography and positron emission tomography, not only enable earlier and more precise detection of disease but also redefine the usual landscape of CAV diagnosis. New methods for individualized risk stratification, including trajectory modeling and machine learning-enhanced biopsy analysis, are paving the way for more personalized surveillance strategies. While current management remains focused on prevention, novel therapeutic targets are emerging, informed by a deeper understanding of CAV immunopathogenesis. This review provides an up-to-date synthesis of recent advances in CAV, with a focus on pathophysiology, individualized risk assessment, diagnostic innovation, and therapeutic perspectives, underscoring a paradigm shift toward more precise and proactive care in heart transplant recipients.

The Alberta Stroke Program Early CT Score (ASPECTS) is widely used to evaluate early ischemic changes and guide thrombectomy decisions in acute stroke patients. However, significant interobserver variability in manual ASPECTS assessment presents a challenge. Recent advances in artificial intelligence have enabled the development of automated ASPECTS scoring systems; however, their comparative performance against expert interpretation remains insufficiently studied. We conducted a systematic review and meta-analysis following PRISMA 2020 guidelines. We searched multiple scientific databases for studies comparing automated and manual ASPECTS on Non-Contrast Computed Tomography (NCCT). Interobserver reliability was assessed using pooled interclass correlation coefficients (ICCs). Subgroup analyses were made using software types, reference standards, time windows, and computed tomography-based factors. Eleven studies with a total of 1,976 patients were included. Automated ASPECTS demonstrated good reliability against reference standards (ICC: 0.72), comparable to expert readings (ICC: 0.62). RAPID ASPECTS performed highest (ICC: 0.86), especially for high-stakes decision-making. AI advantages were most significant with thin-slice CT (≤2.5mm; +0.16), intermediate time windows (120-240min; +0.16), and higher NIHSS scores (p=0.026). AI-driven ASPECTS systems perform comparably or even better in some cases than human readers in detecting early ischemic changes, especially in specific scenarios. Strategic utilization focusing on high-impact scenarios and region-specific performance patterns offers better diagnostic accuracy, reduced interpretation times, and better and wiser treatment selection in acute stroke care.

BackgroundIntracerebral hemorrhage (ICH) is among the most devastating forms of stroke, characterized by high early mortality and limited time-sensitive treatment protocols compared to ischemic stroke. The absence of standardized emergency response frameworks and the shortcomings of conventional scoring systems highlight the urgent need for innovation in neurocritical care. ObjectiveThis paper introduces and evaluates the CODE-ICH framework, along with two AI-powered tools HEADS-UP and SAHVAI designed to transform acute ICH management through real-time detection, volumetric analysis, and predictive modeling. MethodsWe describe the development and implementation of HEADS-UP, a cloud-based AI system for early ICH detection in underserved populations, and SAHVAI, a convolutional neural network-based tool for subarachnoid hemorrhage volume quantification. These tools were integrated into a novel paging and workflow system at a comprehensive stroke center to facilitate ultra-early intervention. ResultsSAHVAI achieved 99.8% accuracy in volumetric analysis and provided 2D, 3D, and 4D visualization of hemorrhage progression. HEADS-UP enabled rapid triage and transfer, reducing reliance on subjective interpretation. Together, these tools operationalized the time is brain principle for hemorrhagic stroke and supported proactive, data-driven care in the neuro-intensive care unit (NICU). ConclusionCODE-ICH, HEADS-UP, and SAHVAI represent a paradigm shift in hemorrhagic stroke care, delivering scalable, explainable, and multimodal AI solutions that enhance clinical decision-making, minimize delays, and promote equitable access to neurocritical care.

Medical Image Segmentation (MIS) stands as a cornerstone in medical image analysis, playing a pivotal role in precise diagnostics, treatment planning, and monitoring of various medical conditions. This paper presents a comprehensive and systematic survey of MIS methodologies, bridging the gap between traditional image processing techniques and modern deep learning approaches. The survey encompasses thresholding, edge detection, region-based segmentation, clustering algorithms, and model-based techniques while also delving into state-of-the-art deep learning architectures such as Convolutional Neural Networks (CNNs), Fully Convolutional Networks (FCNs), and the widely adopted U-Net and its variants. Moreover, integrating attention mechanisms, semi-supervised learning, generative adversarial networks (GANs), and Transformer-based models is thoroughly explored. In addition to covering established methods, this survey highlights emerging trends, including hybrid architectures, cross-modality learning, federated and distributed learning frameworks, and active learning strategies, which aim to address challenges such as limited labeled datasets, computational complexity, and model generalizability across diverse imaging modalities. Furthermore, a specialized case study on lumbar spine segmentation is presented, offering insights into the challenges and advancements in this relatively underexplored anatomical region. Despite significant progress in the field, critical challenges persist, including dataset bias, domain adaptation, interpretability of deep learning models, and integration into real-world clinical workflows.