Accurate interpretation of Magnetic Resonance Imaging scans in clinical systems is based on a precise understanding of image contrast. This contrast is primarily governed by acquisition parameters, such as echo time and repetition time, which are stored in the DICOM metadata. To simplify contrast identification, broad labels such as T1-weighted or T2-weighted are commonly used, but these offer only a coarse approximation of the underlying acquisition settings. In many real-world datasets, such labels are entirely missing, leaving raw acquisition parameters as the only indicators of contrast. Adding to this challenge, the available metadata is often incomplete, noisy, or inconsistent. The lack of reliable and standardized metadata complicates tasks such as image interpretation, retrieval, and integration into clinical workflows. Furthermore, robust contrast-aware representations are essential to enable more advanced clinical applications, such as achieving modality-invariant representations and data harmonization. To address these challenges, we propose MR-CLIP, a multimodal contrastive learning framework that aligns MR images with their DICOM metadata to learn contrast-aware representations, without relying on manual labels. Trained on a diverse clinical dataset that spans various scanners and protocols, MR-CLIP captures contrast variations across acquisitions and within scans, enabling anatomy-invariant representations. We demonstrate its effectiveness in cross-modal retrieval and contrast classification, highlighting its scalability and potential for further clinical applications. The code and weights are publicly available at https://github.com/myigitavci/MR-CLIP.

Magnetic resonance imaging exhibits substantial acquisition variability, where identical anatomy can appear markedly different across scanners and imaging protocols. Consequently, learned representations entangle biological structure with acquisition-dependent appearance, limiting interpretability, generalisation, and clinical deployment. We show that these sources of variation can be separated by jointly modelling MRI images and DICOM metadata. Using large-scale clinical brain MRI data, we learn representations that separate anatomical structure from contrast-dependent appearance. Resulting contrast representations organise heterogeneous acquisitions, support sequence understanding, and detect image–metadata inconsistencies, whereas anatomical representations suppress acquisition-specific variation while preserving biologically relevant information. Building on these disentangled representations, we introduce a unified anatomy-preserving harmonisation model for cross-modality and cross-site adaptation, conditioned on image or acquisition metadata. Our findings suggest that acquisition variability is a structured component of the imaging process that can be modelled, audited, and controlled, providing a foundation for acquisition-aware representation learning in large-scale medical imaging.

Biomedical imaging data exhibit substantial acquisition variability, where identical biological structures can appear markedly different due to differences in imaging devices, acquisition protocols, sites, and reconstruction settings. Consequently, learned representations often entangle underlying biological information with acquisition-dependent appearance, limiting interpretability, generalisation, and clinical deployment. We show that these sources of variation can be separated by jointly modelling medical images and acquisition metadata. Using large-scale clinical brain MRI data as a case study, we learn representations that disentangle anatomical structure from contrast-dependent appearance. The resulting acquisition-aware representations organise heterogeneous imaging protocols, support sequence understanding, and detect image–metadata inconsistencies, whereas anatomical representations suppress acquisition-specific variation while preserving biologically relevant information. Building on these disentangled representations, we introduce a unified anatomy-preserving harmonisation model for cross-modality and cross-site adaptation, conditioned on image content or acquisition metadata. Our findings suggest that acquisition variability is a structured component of the imaging process that can be modelled, audited, and controlled, providing a foundation for acquisition-aware representation learning in large-scale biomedical imaging.

Focal deficits in ischaemic stroke arise primarily from impaired perfusion downstream of a critical vascular occlusion. Though the consequent parenchymal lesion is traditionally used to predict clinical deficits, the underlying pattern of disrupted perfusion provides information upstream of the lesion, potentially yielding earlier predictive and localising signals. We previously developed a technique to compute perfusion maps from routine CT and CT angiography (CTA), an imaging modality widely deployed in clinical practice and available at large data scales. Analysing computed perfusion maps (derived from CT and CTA) from 1393 CTA-imaged patients with confirmed acute ischaemic stroke, here we use deep generative perfusion-deficit inference to localise the neural substrates of NIHSS sub-scores, explicitly disentangling the distinct topologies of disrupted perfusion and neural dependence. We show that our approach replicates known lesion-deficit relations without knowledge of the lesion itself and reveals novel neural dependents. The high achieved anatomical fidelity suggests acute CTA-derived computed perfusion maps may be of substantial clinical and scientific value in rich phenotyping of acute stroke. By relying only on an imaging modality well-established in the hyperacute setting, deep generative perfusion-deficit inference could power highly expressive models of functional anatomical relations in ischaemic stroke within the critical pre-interventional window.

Magnetic Resonance Imaging suffers from substantial data heterogeneity and the absence of standardized contrast labels across scanners, protocols, and institutions, which severely limits large-scale automated analysis. A unified representation of MRI contrast would enable a wide range of downstream utilities, from automatic sequence recognition to harmonization and quality control, without relying on manual annotations. To this end, we introduce MR-CLIP, a metadata-guided framework that learns MRI contrast representations by aligning volumetric images with their DICOM acquisition parameters. The resulting embeddings shows distinct clusters of MRI sequences and outperform supervised 3D baselines under data scarcity in few-shot sequence classification. Moreover, MR-CLIP enables unsupervised data quality control by identifying corrupted or inconsistent metadata through image-metadata embedding distances. By transforming routinely available acquisition metadata into a supervisory signal, MR-CLIP provides a scalable foundation for label-efficient MRI analysis across diverse clinical datasets.

Deep learning holds immense promise for transforming medical image analysis, yet its clinical generalization remains profoundly limited. A major barrier is data heterogeneity. This is particularly true in Magnetic Resonance Imaging, where scanner hardware differences, diverse acquisition protocols, and varying sequence parameters introduce substantial domain shifts that obscure underlying biological signals. Data harmonization methods aim to reduce these instrumental and acquisition variability, but existing approaches remain insufficient. When applied to imaging data, image-based harmonization approaches are often restricted by the need for target images, while existing text-guided methods rely on simplistic labels that fail to capture complex acquisition details or are typically restricted to datasets with limited variability, failing to capture the heterogeneity of real-world clinical environments. To address these limitations, we propose DIST-CLIP (Disentangled Style Transfer with CLIP Guidance), a unified framework for MRI harmonization that flexibly uses either target images or DICOM metadata for guidance. Our framework explicitly disentangles anatomical content from image contrast, with the contrast representations being extracted using pre-trained CLIP encoders. These contrast embeddings are then integrated into the anatomical content via a novel Adaptive Style Transfer module. We trained and evaluated DIST-CLIP on diverse real-world clinical datasets, and showed significant improvements in performance when compared against state-of-the-art methods in both style translation fidelity and anatomical preservation, offering a flexible solution for style transfer and standardizing MRI data. Our code and weights will be made publicly available upon publication.

This study presents a significant advancement in unsupervised magnetic resonance (MR) image harmonization through the development of SR-Param-MPMGen, a novel framework that translates clinical MR acquisitions into an invariant, isotropic space. Building upon existing methods for multi-parametric map (MPM) generation, our approach introduces automatic parameter prediction, enabling physicsbased simulations to convert qualitative MR images into quantitative MPMs without relying on manual parameter inputs or paired training data. The efficacy of SR-Param-MPMGen is validated through image reconstruction-quality analysis and downstream segmentation tasks, demonstrating statistically significant improvements in healthy tissue segmentation compared to a state-of-the-art style transfer method. These results underscore the potential of our approach to improve data harmonization and analysis in large-scale neuroimaging studies and clinical applications, particularly in scenarios with heterogeneous or missing data.

The application of causal discovery to diseases like Alzheimer's (AD) is limited by the static graph assumptions of most methods; such models cannot account for an evolving pathophysiology, modulated by a latent disease pseudotime. We propose to apply an existing latent variable model to real-world AD data, inferring a pseudotime that orders patients along a data-driven disease trajectory independent of chronological age, then learning how causal relationships evolve. Pseudotime outperformed age in predicting diagnosis (AUC 0.82 vs 0.59). Incorporating minimal, disease-agnostic background knowledge substantially improved graph accuracy and orientation. Our framework reveals dynamic interactions between novel (NfL, GFAP) and established AD markers, enabling practical causal discovery despite violated assumptions.

Interactive segmentation is a promising strategy for building robust, generalisable algorithms for volumetric medical image segmentation. However, inconsistent and clinically unrealistic evaluation hinders fair comparison and misrepresents real-world performance. We propose a clinically grounded methodology for defining evaluation tasks and metrics, and built a software framework for constructing standardised evaluation pipelines. We evaluate state-of-the-art algorithms across heterogeneous and complex tasks and observe that (i) minimising information loss when processing user interactions is critical for model robustness, (ii) adaptive-zooming mechanisms boost robustness and speed convergence, (iii) performance drops if validation prompting behaviour/budgets differ from training, (iv) 2D methods perform well with slab-like images and coarse targets, but 3D context helps with large or irregularly shaped targets, (v) performance of non-medical-domain models (e.g. SAM2) degrades with poor contrast and complex shapes.

Medical imaging research is often limited by data scarcity and availability. Governance, privacy concerns and the cost of acquisition all restrict access to medical imaging data, which, compounded by the data-hungry nature of deep learning algorithms, limits progress in the field of healthcare AI. Generative models have recently been used to synthesize photorealistic natural images, presenting a potential solution to the data scarcity problem. But are current generative models synthesizing morphologically correct samples? In this work we present a three-dimensional generative model of the human brain that is trained at the necessary scale to generate diverse, realistic-looking, high-resolution and morphologically preserving samples and conditioned on patient characteristics (for example, age and pathology). We show that the synthetic samples generated by the model preserve biological and disease phenotypes and are realistic enough to permit use downstream in well-established image analysis tools. While the proposed model has broad future applicability, such as anomaly detection and learning under limited data, its generative capabilities can be used to directly mitigate data scarcity, limited data availability and algorithmic fairness. Medical imaging research is limited by data availability. To address this challenge, Tudosiu and colleagues develop a 3D generative model of the human brain that can generate high-resolution morphologically correct brains conditioned on patient characteristics.

Generative modelling and synthetic data can be a surrogate for real medical imaging datasets, whose scarcity and difficulty to share can be a nuisance when delivering accurate deep learning models for healthcare applications. In recent years, there has been an increased interest in using these models for data augmentation and synthetic data sharing, using architectures such as generative adversarial networks (GANs) or diffusion models (DMs). Nonetheless, the application of synthetic data to tasks such as 3D magnetic resonance imaging (MRI) segmentation remains limited due to the lack of labels associated with the generated images. Moreover, many of the proposed generative MRI models lack the ability to generate arbitrary modalities due to the absence of explicit contrast conditioning. These limitations prevent the user from adjusting the contrast and content of the images and obtaining more generalisable data for training task-specific models. In this work, we propose brainSPADE3D, a 3D generative model for brain MRI and associated segmentations, where the user can condition on specific pathological phenotypes and contrasts. The proposed joint imaging-segmentation generative model is shown to generate high-fidelity synthetic images and associated segmentations, with the ability to combine pathologies. We demonstrate how the model can alleviate issues with segmentation model performance when unexpected pathologies are present in the data.

The last few years have seen a boom in using generative models to augment real datasets, as synthetic data can effectively model real data distributions and provide privacy-preserving, shareable datasets that can be used to train deep learning models. However, most of these methods are 2D and provide synthetic datasets that come, at most, with categorical annotations. The generation of paired images and segmentation samples that can be used in downstream, supervised segmentation tasks remains fairly uncharted territory. This work proposes a two-stage generative model capable of producing 2D and 3D semantic label maps and corresponding multi-modal images. We use a latent diffusion model for label synthesis and a VAE-GAN for semantic image synthesis. Synthetic datasets provided by this model are shown to work in a wide variety of segmentation tasks, supporting small, real datasets or fully replacing them while maintaining good performance. We also demonstrate its ability to improve downstream performance on out-of-distribution data.

MRI is a very flexible imaging modality, but with flexibility comes heterogeneity. MRI sequence choice, acquisition parameters, and image resolution form an extrinsic source of variability, reducing our ability to extract the underlying relevant biological signal and causing difficulties in downstream analyses. We propose a new method that can create resolution and acquisition-parameter invariant representations by removing external sources of variability. We use realistic physics models of image resolution and combine them with a differentiable model of MRI sequences to create an invariant high-resolution multi-parametric (MPM) MRI estimate from an arbitrary number of inputs, all trained via self-supervision. The proposed method allows clinical imaging sessions with sequences acquired at arbitrary resolutions to be transformed into a single-domain generalisable representation. We demonstrate the model's validity by showing improved MPM reconstruction and imputation quality compared to previous methods and a significantly improved ability to super-resolve. We also demonstrate domain generalisation capabilities via a downstream classification model that is more robust to the choice of input sequences in an out-of-distribution dataset.

Stroke is a leading cause of disability and death. Effective treatment decisions require early and informative vascular imaging. 4D perfusion imaging is ideal but rarely available within the first hour after stroke, whereas plain CT and CTA usually are. Hence, we propose a framework to extract a predicted perfusion map (PPM) derived from CT and CTA images. In all eighteen patients, we found significantly high spatial similarity (with average Spearman's correlation = 0.7893) between our predicted perfusion map (PPM) and the T-max map derived from 4D-CTP. Voxelwise correlations between the PPM and National Institutes of Health Stroke Scale (NIHSS) subscores for L/R hand motor, gaze, and language on a large cohort of 2,110 subjects reliably mapped symptoms to expected infarct locations. Therefore our PPM could serve as an alternative for 4D perfusion imaging, if the latter is unavailable, to investigate blood perfusion in the first hours after hospital admission.

In the last years, Internet of Things (IoT) platforms have been designed to provide IoT applications with various services such as device discovery, context management, and data filtering. The lack of standardization has led each IoT platform to propose its own abstractions, APIs, and data models. As a consequence, programming interactions between an IoT consuming application and an IoT platform is time-consuming, is error prone, and depends on the developers' level of knowledge about the IoT platform. To address these issues, this article introduces IoTvar, a middleware library deployed on the IoT consumer application that manages all its interactions with IoT platforms. IoTvar relies on declaring variables automatically mapped to sensors whose values are transparently updated with sensor observations through proxies on the client side. This article presents the IoTvar architecture and shows how it has been integrated into the FIWARE, OM2M, and muDEBS platforms. We also report the results of experiments performed to evaluate IoTvar, showing that it reduces the effort required to declare and manage IoT variables and has no considerable impact on CPU, memory, and energy consumption.

Background: Deep learning-based (DL) methods are the best-performing methods for white matter tract seg-mentation in anatomically healthy subjects. However, tract annotations are variable or absent in clinical data and manual annotations are especially difficult in patients with tumors where normal anatomy may be distorted. Direct cortical and subcortical stimulation is the gold standard ground truth to determine the cortical and sub -cortical lo-cation of motor-eloquent areas intra-operatively. Nonetheless, this technique is invasive, prolongs the surgical procedure, and may cause patient fatigue. Navigated Transcranial Magnetic Stimulation (nTMS) has a well-established correlation to direct cortical stimulation for motor mapping and the added advantage of being able to be acquired pre-operatively.New method: In this work, we evaluate the feasibility of using nTMS motor responses as a method to assess corticospinal tract (CST) binary masks and estimated uncertainty generated by a DL-based tract segmentation in patients with diffuse gliomas.Results: Our results show CST binary masks have a high overlap coefficient (OC) with nTMS response masks. A strong negative correlation is found between estimated uncertainty and nTMS response mask distance to the CST binary mask. Comparison with existing methods: We compare our approach (UncSeg) with the state-of-the-art TractSeg in terms of OC between the CST binary masks and nTMS response masks.Conclusions: In this study, we demonstrate that estimated uncertainty from UncSeg is a good measure of the agreement between the CST binary masks and nTMS response masks distance to the CST binary mask boundary.

The optimal expansion of a power system with reduced carbon footprint entails dealing with uncertainty about the distribution of the random variables involved in the decision process. Optimisation under ambiguity sets provides a mechanism to suitably deal with such a setting. For two-stage stochastic linear programs, we propose a new model that is between the optimistic and pessimistic paradigms in distributionally robust stochastic optimisation. When using Wasserstein balls as ambiguity sets, the resulting optimisation problem has nonsmooth convex constraints depending on the number of scenarios and a bilinear objective function. We propose a decomposition method along scenarios that converges to a solution, provided a global optimisation solver for bilinear programs with polyhedral feasible sets is available. The solution procedure is applied to a case study on expansion of energy generation that takes into account sustainability goals for 2050 in Europe, under uncertain future market conditions.

Clinical MR imaging is typically qualitative, i.e. the observed signal reflects the underlying tissue contrast, but the measurements are not meaningful when taken in isolation. Quantitative MR imaging maps are rarely acquired due to time and complexity constraints but directly measure intrinsic tissue properties, allowing for explicit tissue characterisation and MR contrast simulation. A machine learning network trained on quantitative MRI would circumvent the need to design contrast-agnostic models, minimise domain shift issues, and reduce complex data pre-processing. Such models would also be future-proof by design, anticipating new qualitative sequence developments and changes in acquisition parameters. In this work, we propose a new Bloch-equation-based physics-informed unsupervised network that learns to map qualitative MR data to their quantitative equivalents without the need for paired qualitative-quantitative data. Furthermore, we make the proposed model robust to missing data, enabling us to map any arbitrary set of qualitative data from a patient into quantitative Multi-Parametric Maps (MPMs). We demonstrate that the estimated MPMs are a robust and invariant data representation, are self-consistent, enable missing data imputation, and facilitate data harmonisation from multiple sites while bridging algorithmic domain gaps.

Combining multi-site data can strengthen and uncover trends, but is a task that is marred by the influence of site-specific covariates that can bias the data and therefore any downstream analyses. Post-hoc multi-site correction methods exist but have strong assumptions that often do not hold in real-world scenarios. Algorithms should be designed in a way that can account for site-specific effects, such as those that arise from sequence parameter choices, and in instances where generalisation fails, should be able to identify such a failure by means of explicit uncertainty modelling. This body of work showcases such an algorithm, that can become robust to the physics of acquisition in the context of segmentation tasks, while simultaneously modelling uncertainty. We demonstrate that our method not only generalises to complete holdout datasets, preserving segmentation quality, but does so while also accounting for site-specific sequence choices, which also allows it to perform as a harmonisation tool.

The Internet of things (IoT) has been identified as a significant contributor to increasing Information and Communication Technology (ICT) energy consumption in the future. Carefully designing interactions between client applications and systems in the IoT may positively impact energy saving. For this purpose, IoT middleware, the underlying software that manages those interactions, should consider energy efficiency strategies a first-class concern. Furthermore, adding an energy-awareness capability to the middleware could significantly contribute to deepening developers' understanding of energy consumption by applications and helping them minimize the energy demand. In this paper, we propose energy-efficiency strategies and integrate them into an IoT middleware. We also offer a model to calculate the energy consumption of the interactions between an IoT consumer application and an IoT system, which the middleware could use to choose the best strategy to constrain the application's energy consumption to a given budget.