The scarcity of well-annotated diverse medical images is a major hurdle for developing reliable AI models in healthcare. Substantial technical advances have been made in generative foundation models for natural images. Here we develop `ChexGen', a generative vision-language foundation model that introduces a unified framework for text-, mask-, and bounding box-guided synthesis of chest radiographs. Built upon the latent diffusion transformer architecture, ChexGen was pretrained on the largest curated chest X-ray dataset to date, consisting of 960,000 radiograph-report pairs. ChexGen achieves accurate synthesis of radiographs through expert evaluations and quantitative metrics. We demonstrate the utility of ChexGen for training data augmentation and supervised pretraining, which led to performance improvements across disease classification, detection, and segmentation tasks using a small fraction of training data. Further, our model enables the creation of diverse patient cohorts that enhance model fairness by detecting and mitigating demographic biases. Our study supports the transformative role of generative foundation models in building more accurate, data-efficient, and equitable medical AI systems.

Collimation in X-ray imaging restricts exposure to the region-of-interest (ROI) and minimizes the radiation dose applied to the patient. The detection of collimator shadows is an essential image-based preprocessing step in digital radiography posing a challenge when edges get obscured by scattered X-ray radiation. Regardless, the prior knowledge that collimation forms polygonal-shaped shadows is evident. For this reason, we introduce a deep learning-based segmentation that is inherently constrained to its geometry. We achieve this by incorporating a differentiable Hough transform-based network to detect the collimation borders and enhance its capability to extract the information about the ROI center. During inference, we combine the information of both tasks to enable the generation of refined, line-constrained segmentation masks. We demonstrate robust reconstruction of collimated regions achieving median Hausdorff distances of 4.3-5.0mm on diverse test sets of real Xray images. While this application involves at most four shadow borders, our method is not fundamentally limited by a specific number of edges.

Accurate quantification of tau pathology via tau positron emission tomography (PET) scan is crucial for diagnosing and monitoring Alzheimer's disease (AD). However, the high cost and limited availability of tau PET restrict its widespread use. In contrast, structural magnetic resonance imaging (MRI) and plasma-based biomarkers provide non-invasive and widely available complementary information related to brain anatomy and disease progression. In this work, we propose a text-guided 3D diffusion model for 3D tau PET image synthesis, leveraging multimodal conditions from both structural MRI and plasma measurement. Specifically, the textual prompt is from the plasma p-tau217 measurement, which is a key indicator of AD progression, while MRI provides anatomical structure constraints. The proposed framework is trained and evaluated using clinical AV1451 tau PET data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. Experimental results demonstrate that our approach can generate realistic, clinically meaningful 3D tau PET across a range of disease stages. The proposed framework can help perform tau PET data augmentation under different settings, provide a non-invasive, cost-effective alternative for visualizing tau pathology, and support the simulation of disease progression under varying plasma biomarker levels and cognitive conditions.

Medical foundation models (FMs) have shown tremendous promise amid the rapid advancements in artificial intelligence (AI) technologies. However, current medical FMs typically generate answers in a black-box manner, lacking transparent reasoning processes and locally grounded interpretability, which hinders their practical clinical deployments. To this end, we introduce DeepMedix-R1, a holistic medical FM for chest X-ray (CXR) interpretation. It leverages a sequential training pipeline: initially fine-tuned on curated CXR instruction data to equip with fundamental CXR interpretation capabilities, then exposed to high-quality synthetic reasoning samples to enable cold-start reasoning, and finally refined via online reinforcement learning to enhance both grounded reasoning quality and generation performance. Thus, the model produces both an answer and reasoning steps tied to the image's local regions for each query. Quantitative evaluation demonstrates substantial improvements in report generation (e.g., 14.54% and 31.32% over LLaVA-Rad and MedGemma) and visual question answering (e.g., 57.75% and 23.06% over MedGemma and CheXagent) tasks. To facilitate robust assessment, we propose Report Arena, a benchmarking framework using advanced language models to evaluate answer quality, further highlighting the superiority of DeepMedix-R1. Expert review of generated reasoning steps reveals greater interpretability and clinical plausibility compared to the established Qwen2.5-VL-7B model (0.7416 vs. 0.2584 overall preference). Collectively, our work advances medical FM development toward holistic, transparent, and clinically actionable modeling for CXR interpretation.

This paper develops dimension reduction techniques for accelerating diffusion model inference in the context of synthetic data generation. The idea is to integrate compressed sensing into diffusion models: (i) compress the data into a latent space, (ii) train a diffusion model in the latent space, and (iii) apply a compressed sensing algorithm to the samples generated in the latent space, facilitating the efficiency of both model training and inference. Under suitable sparsity assumptions on data, the proposed algorithm is proved to enjoy faster convergence by combining diffusion model inference with sparse recovery. As a byproduct, we obtain an optimal value for the latent space dimension. We also conduct numerical experiments on a range of datasets, including image data (handwritten digits, medical images, and climate data) and financial time series for stress testing.

Threatened miscarriage (TM), defined as first-trimester vaginal bleeding with a closed cervix and detectable fetal cardiac activity, affects up to 30 % of clinically recognised pregnancies and is linked to increased risk of adverse outcomes. This study evaluates the predictive value of first-trimester ultrasound (US) and biochemical (BC) markers in determining outcomes among women with TM symptoms. This prospective cohort study recruited 118 women with viable singleton pregnancies (5⁺⁰ to 12⁺⁶ weeks' gestation) from Malta's national public hospital between January 2023 and June 2024. Participants underwent US and BC assessment, along with collection of clinical and sociodemographic data. Pregnancy outcomes were followed to term and classified as live birth or loss. Univariate logistic regression identified individual predictors. Multivariate logistic regression (MLR) and random forest (RF) modelling assessed combined predictive performance. Among 118 TM cases, 77 % resulted in live birth, 23 % in loss. MLR identified progesterone, cervical length, mean gestational sac diameter (MGSD), trophoblast thickness, sFlt-1:PlGF ratio, and maternal age as significant predictors. Higher progesterone, cervical length, MGSD, and sFlt-1:PlGF ratio reduced risk, while maternal age over 35 increased it. MLR achieved 82.7 % accuracy (AUC = 0.89). RF improved accuracy to 93.1 % (AUC = 0.97), confirming the combined predictive value of US and BC markers. US and BC markers hold predictive value in TM. Machine learning, particularly RF, may improve early clinical risk stratification. This tool may support timely decision-making and personalised monitoring, intervention, and counselling for women with TM.

Non-contrast computed tomography (NCCT) is a first-line imaging technique for determining treatment options for acute ischemic stroke (AIS). However, its poor contrast and signal-to-noise ratio limit the diagnosis accuracy for radiologists, and automated AIS lesion segmentation using NCCT also remains a challenge. This study aims to develop a segmentation method for ischemic lesions in NCCT scans, combining symmetry-based principles with the nnUNet segmentation model. Our novel approach integrates a Generative Module (GM) utilizing 2.5 D ResUNet and an Upstream Segmentation Module (UM) with additional inputs and constraints under the 3D nnUNet segmentation model, utilizing symmetry-based learning to enhance the identification and segmentation of ischemic regions. We utilized the publicly accessible AISD dataset for our experiments. This dataset contains 397 NCCT scans of acute ischemic stroke taken within 24 h of the onset of symptoms. Our method was trained and validated using 345 scans, while the remaining 52 scans were used for internal testing. Additionally, we included 60 positive cases (External Set 1) with segmentation labels obtained from our hospital for external validation of the segmentation task. External Set 2 was employed to evaluate the model's sensitivity and specificity in case-dimensional classification, further assessing its clinical performance. We introduced innovative features such as an intensity-based lesion probability (ILP) function and specific input channels for suspected lesion areas to augment the model's sensitivity and specificity. The methodology demonstrated commendable segmentation efficacy, attaining a Dice Similarity Coefficient (DSC) of 0.6720 and a Hausdorff Distance (HD95) of 35.28 on the internal test dataset. Similarly, on the external test dataset, the method yielded satisfactory segmentation outcomes, with a DSC of 0.4891 and an HD 95 of 46.06. These metrics reflect a substantial overlap with expert-drawn boundaries and demonstrate the model's potential for reliable clinical application. In terms of classification performance, the method achieved an Area Under the Curve (AUC) of 0.991 on the external test set, surpassing the performance of nnUNet, which recorded an AUC of 0.947. This study introduces a novel segmentation technique for ischemic lesions in NCCT scans, leveraging symmetry-based principles integrated with nnUNet, which shows potential for improving clinical decision-making in stroke care.

Lung lobe segmentation is required to assess lobar function with nuclear imaging before surgical interventions. We evaluated the performance of open-source deep learning-based lung lobe segmentation tools, compared to a similar nnU-Net model trained on a smaller but more representative clinical dataset. We collated and semi-automatically segmented an internal dataset of 164 computed tomography scans and classified them for task difficulty as easy, moderate, or hard. The performance of three open-source models-multi-organ objective segmentation (MOOSE), TotalSegmentator, and LungMask-was assessed using Dice similarity coefficient (DSC), robust Hausdorff distance (rHd95), and normalized surface distance (NSD). Additionally, we trained, validated, and tested an nnU-Net model using our local dataset and compared its performance with that of the other software on the test subset. All models were evaluated for generalizability using an external competition (LOLA11, n = 55). TotalSegmentator outperformed MOOSE in DSC and NSD across all difficulty levels (p < 0.001), but not in rHd95 (p = 1.000). MOOSE and TotalSegmentator surpassed LungMask across metrics and difficulty classes (p < 0.001). Our model exceeded all other models on the internal dataset (n = 33) in all metrics, across all difficulty classes (p < 0.001), and on the external dataset. Missing lobes were correctly identified only by our model and LungMask in 3 and 1 of 7 cases, respectively. Open-source segmentation tools perform well in straightforward cases but struggle in unfamiliar, complex cases. Training on diverse, specialized datasets can improve generalizability, emphasizing representative data over sheer quantity. Training lung lobe segmentation models on a local variety of cases improves accuracy, thus enhancing presurgical planning, ventilation-perfusion analysis, and disease localization, potentially impacting treatment decisions and patient outcomes in respiratory and thoracic care. Deep learning models trained on non-specialized datasets struggle with complex lung anomalies, yet their real-world limitations are insufficiently assessed. Training an identical model on a smaller yet clinically diverse and representative cohort improved performance in challenging cases. Data diversity outweighs the quantity in deep learning-based segmentation models. Accurate lung lobe segmentation may enhance presurgical assessment of lung lobar ventilation and perfusion function, optimizing clinical decision-making and patient outcomes.

To develop a cascaded deep learning (DL) framework integrating tumor segmentation with metastatic risk stratification for preoperative prediction of occult peritoneal metastasis (OPM) in advanced gastric cancer (GC), and validate its generalizability for early peritoneal recurrence (PR) prediction. This multicenter study enrolled 765 patients with advanced GC from three institutions. We developed a two-stage framework as follows: (1) V-Net-based tumor segmentation on CT; (2) DL-based metastatic risk classification using segmented tumor regions. Clinicopathological predictors were integrated with deep learning probabilities to construct a combined model. Validation cohorts comprised: Internal validation (Test1 for OPM, n=168; Test2 for early PR, n=212) and External validation (Test3 for early PR, n=57 from two independent centers). Multivariable analysis identified Borrmann type (OR=1.314, 95% CI: 1.239-1.394), CA125 ≥35U/mL (OR=1.301, 95% CI: 1.127-1.499), and CT-N+ stage (OR=1.259, 95% CI: 1.124-1.415) as independent OPM predictors. The combined model demonstrated robust performance for both OPM and early PR prediction: achieving AUCs of 0.938 (Train) and 0.916 (Test1) for OPM with improvements over clinical (∆AUC +0.039-+0.107) and DL-only models (∆AUC +0.044-+0.104), while attaining AUC 0.820-0.825 for early PR (Test2 and Test3) with balanced sensitivity (79.7-88.9%) and specificity (72.4-73.3%). Decision curve analysis confirmed net clinical benefit across clinical thresholds. This CT-based cascaded framework enables reliable preoperative risk stratification for OPM and early PR in advanced GC, potentially refining indications for personalized therapeutic pathways.

To develop deep learning-based multiomics models for predicting postoperative distant metastasis (DM) and evaluating survival prognosis in colorectal cancer (CRC) patients. This retrospective study included 521 CRC patients who underwent curative surgery at two centers. Preoperative CT and postoperative hematoxylin-eosin (HE) stained slides were collected. A total of 381 patients from Center 1 were split (7:3) into training and internal validation sets; 140 patients from Center 2 formed the independent external validation set. Patients were grouped based on DM status during follow-up. Radiological and pathological models were constructed using independent imaging and pathological predictors. Deep features were extracted with a ResNet-101 backbone to build deep learning radiomics (DLRS) and deep learning pathomics (DLPS) models. Two integrated models were developed: Nomogram 1 (radiological + DLRS) and Nomogram 2 (pathological + DLPS). CT- reported T (cT) stage (OR=2.00, P=0.006) and CT-reported N (cN) stage (OR=1.63, P=0.023) were identified as independent radiologic predictors for building the radiological model; pN stage (OR=1.91, P=0.003) and perineural invasion (OR=2.07, P=0.030) were identified as pathological predictors for building the pathological model. DLRS and DLPS incorporated 28 and 30 deep features, respectively. In the training set, area under the curve (AUC) for radiological, pathological, DLRS, DLPS, Nomogram 1, and Nomogram 2 models were 0.657, 0.687, 0.931, 0.914, 0.938, and 0.930. DeLong's test showed DLRS, DLPS, and both nomograms significantly outperformed conventional models (P<.05). Kaplan-Meier analysis confirmed effective 3-year disease-free survival (DFS) stratification by the nomograms. Deep learning-based multiomics models provided high accuracy for postoperative DM prediction. Nomogram models enabled reliable DFS risk stratification in CRC patients.