Fat fraction (FF) quantification in individual muscles using quantitative MRI is of major importance for monitoring disease progression and assessing disease severity in neuromuscular diseases. Undersampling of MRI acquisitions is commonly used to reduce scanning time. The present paper introduces novel unrolled neural networks for the reconstruction of undersampled MRI acquisitions. These networks are designed with the aim of maintaining accurate FF quantification while reducing reconstruction time and memory usage. The proposed approach relies on a combination of a simplified architecture (Half U-Net) with unrolled networks that achieved high performance in the well-known FastMRI challenge (variational network [VarNet] and densely interconnected residual cascading network [DIRCN]). The algorithms were trained and evaluated using 3D MRI Dixon acquisitions of the thigh from controls and patients with neuromuscular diseases. The study was performed by applying a retrospective undersampling with acceleration factors of 4 and 8. Reconstructed images were used to computed FF maps. Results disclose that the novel unrolled neural networks were able to maintain reconstruction, biomarker assessment, and segmentation quality while reducing memory usage by 24% to 16% and reducing reconstruction time from 21% to 17%. Using an acceleration factor of 8, the proposed algorithms, HalfVarNet and HalfDIRCN, achieved structural similarity index (SSIM) scores of 93.76 ± 0.38 and 94.95 ± 0.32, mean squared error (MSE) values of 12.76 ± 1.08 × 10^-2 and 10.25 ± 0.87 × 10^-2, and a relative FF quadratic error of 0.23 ± 0.02% and 0.17 ± 0.02%, respectively. The proposed method enables time and memory-efficient reconstruction of undersampled 3D MRI data, supporting its potential for clinical application.

AI-assisted radiological interpretation is based on predominantly narrow, single-task models. This approach is impractical for covering the vast spectrum of imaging modalities, diseases, and radiological findings. Foundation models (FMs) hold the promise of broad generalization across modalities and in low-data settings. However, this potential has remained largely unrealized in radiology. We introduce Curia, a foundation model trained on the entire cross-sectional imaging output of a major hospital over several years, which to our knowledge is the largest such corpus of real-world data-encompassing 150,000 exams (130 TB). On a newly curated 19-task external validation benchmark, Curia accurately identifies organs, detects conditions like brain hemorrhages and myocardial infarctions, and predicts outcomes in tumor staging. Curia meets or surpasses the performance of radiologists and recent foundation models, and exhibits clinically significant emergent properties in cross-modality, and low-data regimes. To accelerate progress, we release our base model's weights at https://huggingface.co/raidium/curia.

BackgroundCardiac allograft rejection (CAR) remains the leading cause of early graft failure after heart transplantation (HT). Current diagnostics, including histologic grading of endomyocardial biopsy (EMB) and blood-based assays, lack accurate predictive power for future CAR risk. We developed a predictive model integrating routine clinical data with quantitative morphologic features extracted from routine EMBs to demonstrate the precision-medicine potential of mining existing data sources in post-HT care. MethodsIn a retrospective cohort of 484 HT recipients with 1,188 EMB encounters within 6 months post-transplant, we extracted 370 quantitative pathology features describing lymphocyte infiltration and stromal architecture from digitized H&E-stained slides. Longitudinal clinical data comprising 268 variables--including lab values, immunosuppression records, and prior rejection history--were aggregated per patient. Using the XGBoost algorithm with rigorous cross-validation, we compared models based on four different data sources: clinical-only, morphology-only, cross-sectional-only, and fully integrated longitudinal data. The top predictors informed the derivation of a simplified Integrated Rejection Risk Index (IRRI), which relies on just 4 clinical and 4 morphology risk facts. Model performance was evaluated by AUROC, AUPRC, and time-to-event hazard ratios. ResultsThe fully integrated longitudinal model achieved superior predictive accuracy (AUROC 0.86, AUPRC 0.74). IRRI stratified patients into risk categories with distinct future CAR hazards: high-risk patients showed a markedly increased CAR risk (HR=6.15, 95% CI: 4.17-9.09), while low-risk patients had significantly reduced risk (HR=0.52, 95% CI: 0.33-0.84). This performance exceeded models based on just cross-sectional or single-domain data, demonstrating the value of multi-modal, temporal data integration. ConclusionsBy integrating longitudinal clinical and biopsy morphologic features, IRRI provides a scalable, interpretable tool for proactive CAR risk assessment. This precision-based approach could support risk-adaptive surveillance and immunosuppression management strategies, offering a promising pathway toward safer, more personalized post-HT care with the potential to reduce unnecessary procedures and improve outcomes. Clinical PerspectiveWhat is new? O_LICurrent tools for cardiac allograft monitoring detect rejection only after it occurs and are not designed to forecast future risk. This leads to missed opportunities for early intervention, avoidable patient injury, unnecessary testing, and inefficiencies in care. C_LIO_LIWe developed a machine learning-based risk index that integrates clinical features, quantitative biopsy morphology, and longitudinal temporal trends to create a robust predictive framework. C_LIO_LIThe Integrated Rejection Risk Index (IRRI) provides highly accurate prediction of future allograft rejection, identifying both high- and low-risk patients up to 90 days in advance - a capability entirely absent from current transplant management. C_LI What are the clinical implications? O_LIIntegrating quantitative histopathology with clinical data provides a more precise, individualized estimate of rejection risk in heart transplant recipients. C_LIO_LIThis framework has the potential to guide post-transplant surveillance intensity, immunosuppressive management, and patient counseling. C_LIO_LIAutomated biopsy analysis could be incorporated into digital pathology workflows, enabling scalable, multicenter application in real-world transplant care. C_LI

Preterm birth is a major cause of mortality and lifelong morbidity in childhood. Its complex and multifactorial origins limit the effectiveness of current clinical predictors and impede optimal care. In this study, a dual-branch deep learning architecture (PUUMA) was developed to predict gestational age (GA) at birth using T2* fetal MRI data from 295 pregnancies, encompassing a heterogeneous and imbalanced population. The model integrates both global whole-uterus and local placental features. Its performance was benchmarked against linear regression using cervical length measurements obtained by experienced clinicians from anatomical MRI and other Deep Learning architectures. The GA at birth predictions were assessed using mean absolute error. Accuracy, sensitivity, and specificity were used to assess preterm classification. Both the fully automated MRI-based pipeline and the cervical length regression achieved comparable mean absolute errors (3 weeks) and good sensitivity (0.67) for detecting preterm birth, despite pronounced class imbalance in the dataset. These results provide a proof of concept for automated prediction of GA at birth from functional MRI, and underscore the value of whole-uterus functional imaging in identifying at-risk pregnancies. Additionally, we demonstrate that manual, high-definition cervical length measurements derived from MRI, not currently routine in clinical practice, offer valuable predictive information. Future work will focus on expanding the cohort size and incorporating additional organ-specific imaging to improve generalisability and predictive performance.

PurposeTo develop and externally validate a multimodal AI model for detecting ischaemia complicating small-bowel obstruction (SBO). MethodsWe combined 3D CT data with routine laboratory markers (C-reactive protein, neutrophil count) and, optionally, radiology report text. From two centers, 1,350 CT examinations were curated; 771 confirmed SBO scans were used for model development with patient-level splits. Ischemia labels were defined by surgical confirmation within 24 hours of imaging. Models (MViT, ResNet-101, DaViT) were trained as unimodal and multimodal variants. External testing was used for 66 independent cases from a third center. Two radiologists (attending, resident) read the test set with and without AI assistance. Performance was assessed using AUC, sensitivity, specificity, and 95% bootstrap confidence intervals; predictions included a confidence score. ResultsThe image-plus-laboratory model performed best on external testing (AUC 0.69 [0.59-0.79], sensitivity 0.89 [0.76-1.00], and specificity 0.44 [0.35-0.54]). Adding report text improved internal validation but did not generalize externally; image+text and full multimodal variants did not exceed image+laboratory performance. Without AI, the attending outperformed the resident (AUC 0.745 [0.617-0.845] vs 0.706 [0.581-0.818]); with AI, both improved, attending 0.752 [0.637-0.853] and resident 0.752 [0.629-0.867], rising to 0.750 [0.631-0.839] and 0.773 [0.657-0.867] with confidence display; differences were not statistically significant. ConclusionA multimodal AI that combines CT images with routine laboratory markers outperforms single-modality approaches and boosts radiologist readers performance notably junior, supporting earlier, more consistent decisions within the first 24 hours. Key PointsA multimodal artificial intelligence (AI) model that combines CT images with laboratory markers detected ischemia in small-bowel obstruction with AUC 0.69 (95% CI 0.59-0.79) and sensitivity 0.89 (0.76-1.00) on external testing, outperforming single-modality models. Adding report text did not generalize across sites: the image+text model fell from AUC 0.82 (internal) to 0.53 (external), and adding text to image+biology left external AUC unchanged (0.69) with similar specificity (0.43-0.44). With AI assistance both junior and senior readers improved; the juniors AUC rose from 0.71 to 0.77, reaching senior-level performance. Summary StatementA multicentric AI model combining CT and routine laboratory data (CRP and neutrophilia) improved radiologists detection of ischemia in small-bowel obstruction. This tool supports earlier decision-making within the first 24 hours.

Background: Precise breast ultrasound (BUS) segmentation supports reliable measurement, quantitative analysis, and downstream classification, yet remains difficult for small or low-contrast lesions with fuzzy margins and speckle noise. Text prompts can add clinical context, but directly applying weakly localized text-image cues (e.g., CAM/CLIP-derived signals) tends to produce coarse, blob-like responses that smear boundaries unless additional mechanisms recover fine edges. Methods: We propose XBusNet, a novel dual-prompt, dual-branch multimodal model that combines image features with clinically grounded text. A global pathway based on a CLIP Vision Transformer encodes whole-image semantics conditioned on lesion size and location, while a local U-Net pathway emphasizes precise boundaries and is modulated by prompts that describe shape, margin, and Breast Imaging Reporting and Data System (BI-RADS) terms. Prompts are assembled automatically from structured metadata, requiring no manual clicks. We evaluate on the Breast Lesions USG (BLU) dataset using five-fold cross-validation. Primary metrics are Dice and Intersection over Union (IoU); we also conduct size-stratified analyses and ablations to assess the roles of the global and local paths and the text-driven modulation. Results: XBusNet achieves state-of-the-art performance on BLU, with mean Dice of 0.8765 and IoU of 0.8149, outperforming six strong baselines. Small lesions show the largest gains, with fewer missed regions and fewer spurious activations. Ablation studies show complementary contributions of global context, local boundary modeling, and prompt-based modulation. Conclusions: A dual-prompt, dual-branch multimodal design that merges global semantics with local precision yields accurate BUS segmentation masks and improves robustness for small, low-contrast lesions.

PurposeTo evaluate convolutional neural network (CNN) model training strategies that optimize the performance of calcaneus fracture detection on radiographs at different image resolutions. Materials and MethodsThis retrospective study included foot radiographs from a single hospital between 2015 and 2022 for a total of 1,775 x-ray series (551 fractures; 1,224 without) and was split into training (70%), validation (15%), and testing (15%). ImageNet pre-trained ResNet models were fine-tuned on the dataset. Three training strategies were evaluated: 1) single size: trained exclusively on 128x128, 256x256, 512x512, 640x640, or 900x900 radiographs (5 model sets); 2) curriculum learning: trained exclusively on 128x128 radiographs then exclusively on 256x256, then 512x512, then 640x640, and finally on 900x900 (5 model sets); and 3) multi-scale augmentation: trained on x-ray images resized along continuous dimensions between 128x128 to 900x900 (1 model set). Inference time and training time were compared. ResultsMulti-scale augmentation trained models achieved the highest average area under the Receiver Operating Characteristic curve of 0.938 [95% CI: 0.936 - 0.939] for a single model across image resolutions compared to the other strategies without prolonging training or inference time. Using the optimal model sets, curriculum learning had the highest sensitivity on in-distribution low-resolution images (85.4% to 90.1%) and on out-of-distribution high-resolution images (78.2% to 89.2%). However, curriculum learning models took significantly longer to train (11.8 [IQR: 11.1-16.4] hours; P<.001). ConclusioWhile 512x512 images worked well for fracture identification, curriculum learning and multi-scale augmentation training strategies algorithmically improved model robustness towards different image resolutions without requiring additional annotated data. Summary statementDifferent deep learning training strategies affect performance in detecting calcaneus fractures on radiographs across in- and out-of-distribution image resolutions, with a multi-scale augmentation strategy conferring the greatest overall performance improvement in a single model. Key pointsO_LITraining strategies addressing differences in radiograph image resolution (or pixel dimensions) could improve deep learning performance. C_LIO_LIThe highest average performance across different image resolutions in a single model was achieved by multi-scale augmentation, where the sampled training dataset is uniformly resized between square resolutions of 128x128 to 900x900. C_LIO_LICompared to model training on a single image resolution, sequentially training on increasingly higher resolution images up to 900x900 (i.e., curriculum learning) resulted in higher fracture detection performance on images resolutions between 128x128 and 2048x2048. C_LI

Automated cardiac MR segmentation enables accurate and reproducible ventricular function assessment in Tetralogy of Fallot (ToF), whereas manual segmentation remains time-consuming and variable. To evaluate the deep learning (DL)-based models for automatic left ventricle (LV), right ventricle (RV), and LV myocardium segmentation in ToF, compared with manual reference standard annotations. Retrospective. 427 patients with diverse cardiac conditions (305 non-ToF, 122 ToF), with 395 for training/validation, 32 ToF for internal testing, and 12 external ToF for generalizability assessment. Steady-state free precession cine sequence at 1.5/3 T. U-Net, Deep U-Net, and MultiResUNet were trained under three regimes (non-ToF, ToF-only, mixed), using manual segmentations from one radiologist and one researcher (20 and 10 years of experience, respectively) as reference, with consensus for discrepancies. Performance for LV, RV, and LV myocardium was evaluated using Dice Similarity Coefficient (DSC), Intersection over Union (IoU), and F1-score, alongside regional (basal, middle, apical) and global ventricular function comparisons to manual results. Friedman tests were applied for architecture and regime comparisons, paired Wilcoxon tests for ED-ES differences, and Pearson's r for assessing agreement in global function. MultiResUNet model trained on a mixed dataset (TOF and non-TOF cases) achieved the best segmentation performance, with DSCs of 96.1% for LV and 93.5% for RV. In the internal test set, DSCs for LV, RV, and LV myocardium were 97.3%, 94.7%, and 90.7% at end-diastole, and 93.6%, 92.1%, and 87.8% at end-systole, with ventricular measurement correlations ranging from 0.84 to 0.99. Regional analysis showed LV DSCs of 96.3% (basal), 96.4% (middle), and 94.1% (apical), and RV DSCs of 92.8%, 94.2%, and 89.6%. External validation (n = 12) showed correlations ranging from 0.81 to 0.98. The MultiResUNet model enabled accurate automated cardiac MRI segmentation in ToF with the potential to streamline workflows and improve disease monitoring. 3. Stage 2.

Diffusion MRI (dMRI) angular super-resolution (ASR) aims to reconstruct high-angular-resolution (HAR) signals from limited low-angular-resolution (LAR) data without prolonging scan time. However, existing methods are limited in recovering fine-grained angular details or preserving high fidelity due to inadequate modeling of q-space geometry and insufficient incorporation of physical constraints. In this paper, we introduce a Physics-Guided Diffusion Transformer (PGDiT) designed to explore physical priors throughout both training and inference stages. During training, a Q-space Geometry-Aware Module (QGAM) with b-vector modulation and random angular masking facilitates direction-aware representation learning, enabling the network to generate directionally consistent reconstructions with fine angular details from sparse and noisy data. In inference, a two-stage Spherical Harmonics-Guided Posterior Sampling (SHPS) enforces alignment with the acquired data, followed by heat-diffusion-based SH regularization to ensure physically plausible reconstructions. This coarse-to-fine refinement strategy mitigates oversmoothing and artifacts commonly observed in purely data-driven or generative models. Extensive experiments on general ASR tasks and two downstream applications, Diffusion Tensor Imaging (DTI) and Neurite Orientation Dispersion and Density Imaging (NODDI), demonstrate that PGDiT outperforms existing deep learning models in detail recovery and data fidelity. Our approach presents a novel generative ASR framework that offers high-fidelity HAR dMRI reconstructions, with potential applications in neuroscience and clinical research.

Choledochal cysts (CC) and cystic biliary atresia (CBA) present similarly in early infancy but require different treatment approaches. While CC surgery can be delayed until 3-6 months of age in asymptomatic patients, CBA requires intervention within 60 days to prevent cirrhosis. To develop a diagnostic model for early differentiation between these conditions. A total of 319 patients with hepatic hilar cysts (< 60 days old at surgery) were retrospectively analyzed; these patients were treated at three hospitals between 2011 and 2022. Clinical features including biochemical markers and ultrasonographic measurements were compared between CC (<i>n</i> = 274) and CBA (<i>n</i> = 45) groups. Least absolute shrinkage and selection operator regression identified key diagnostic features, and 11 machine learning models were developed and compared. The CBA group showed higher levels of total bile acid, total bilirubin, γ-glutamyl transferase, aspartate aminotransferase, and alanine aminotransferase, and direct bilirubin, while longitudinal diameter of the cysts and transverse diameter of the cysts were larger in the CC group. The multilayer perceptron model demonstrated optimal performance with 95.8% accuracy, 92.9% sensitivity, 96.3% specificity, and an area under the curve of 0.990. Decision curve analysis confirmed its clinical utility. Based on the model, we developed user-friendly diagnostic software for clinical implementation. Our machine learning approach differentiates CC from CBA in early infancy using routinely available clinical parameters. Early accurate diagnosis facilitates timely surgical intervention for CBA cases, potentially improving patient outcomes.