Low-dose computed tomography (CT) denoising is crucial for reduced radiation exposure while ensuring diagnostically acceptable image quality. Despite significant advancements driven by deep learning (DL) in recent years, existing DL-based methods, typically trained on a specific dose level and anatomical region, struggle to handle diverse noise characteristics and anatomical heterogeneity during varied scanning conditions, limiting their generalizability and robustness in clinical scenarios. In this paper, we propose FoundDiff, a foundational diffusion model for unified and generalizable LDCT denoising across various dose levels and anatomical regions. FoundDiff employs a two-stage strategy: (i) dose-anatomy perception and (ii) adaptive denoising. First, we develop a dose- and anatomy-aware contrastive language image pre-training model (DA-CLIP) to achieve robust dose and anatomy perception by leveraging specialized contrastive learning strategies to learn continuous representations that quantify ordinal dose variations and identify salient anatomical regions. Second, we design a dose- and anatomy-aware diffusion model (DA-Diff) to perform adaptive and generalizable denoising by synergistically integrating the learned dose and anatomy embeddings from DACLIP into diffusion process via a novel dose and anatomy conditional block (DACB) based on Mamba. Extensive experiments on two public LDCT datasets encompassing eight dose levels and three anatomical regions demonstrate superior denoising performance of FoundDiff over existing state-of-the-art methods and the remarkable generalization to unseen dose levels. The codes and models are available at https://github.com/hao1635/FoundDiff.

In this study, as a proof-of-concept, we aim to initiate the development of Radiology Foundation Model, termed as RadFM. We consider three perspectives: dataset construction, model design, and thorough evaluation, concluded as follows: (i), we contribute 4 multimodal datasets with 13M 2D images and 615K 3D scans. When combined with a vast collection of existing datasets, this forms our training dataset, termed as Medical Multi-modal Dataset, MedMD. (ii), we propose an architecture that enables to integrate text input with 2D or 3D medical scans, and generates responses for diverse radiologic tasks, including diagnosis, visual question answering, report generation, and rationale diagnosis; (iii), beyond evaluation on 9 existing datasets, we propose a new benchmark, RadBench, comprising three tasks aiming to assess foundation models comprehensively. We conduct both automatic and human evaluations on RadBench. RadFM outperforms former accessible multi-modal foundation models, including GPT-4V. Additionally, we adapt RadFM for diverse public benchmarks, surpassing various existing SOTAs.

Proper segmentation of organs-at-risk is important for radiation therapy, surgical planning, and diagnostic decision-making in medical image analysis. While deep learning-based segmentation architectures have made significant progress, they often fail to balance segmentation accuracy with computational efficiency. Most of the current state-of-the-art methods either prioritize performance at the cost of high computational complexity or compromise accuracy for efficiency. This paper addresses this gap by introducing an efficient dual-line decoder segmentation network (EDLDNet). The proposed method features a noisy decoder, which learns to incorporate structured perturbation at training time for better model robustness, yet at inference time only the noise-free decoder is executed, leading to lower computational cost. Multi-Scale convolutional Attention Modules (MSCAMs), Attention Gates (AGs), and Up-Convolution Blocks (UCBs) are further utilized to optimize feature representation and boost segmentation performance. By leveraging multi-scale segmentation masks from both decoders, we also utilize a mutation-based loss function to enhance the model's generalization. Our approach outperforms SOTA segmentation architectures on four publicly available medical imaging datasets. EDLDNet achieves SOTA performance with an 84.00% Dice score on the Synapse dataset, surpassing baseline model like UNet by 13.89% in Dice score while significantly reducing Multiply-Accumulate Operations (MACs) by 89.7%. Compared to recent approaches like EMCAD, our EDLDNet not only achieves higher Dice score but also maintains comparable computational efficiency. The outstanding performance across diverse datasets establishes EDLDNet's strong generalization, computational efficiency, and robustness. The source code, pre-processed data, and pre-trained weights will be available at https://github.com/riadhassan/EDLDNet .

Addressing the critical challenge of precise boundary delineation in medical image segmentation, we introduce DPGNet, an adaptive deep learning model engineered to emulate expert perception of intricate anatomical edges. Our key innovations drive its superior performance and clinical utility, encompassing: 1) a three-stage progressive refinement strategy that establishes global context, performs hierarchical feature enhancement, and precisely delineates local boundaries; 2) a novel Edge Difference Attention (EDA) module that implicitly learns and quantifies boundary uncertainties without requiring explicit ground truth supervision; and 3) a lightweight, transformer-based architecture ensuring an exceptional balance between performance and computational efficiency. Extensive experiments across diverse and challenging medical image datasets demonstrate DPGNet's consistent superiority over state-of-the-art methods, notably achieving this with significantly lower computational overhead (25.51 M parameters). Its exceptional boundary refinement is rigorously validated through comprehensive metrics (Boundary-IoU, HD95) and confirmed by rigorous clinical expert evaluations. Crucially, DPGNet generates an explicit uncertainty boundary map, providing clinicians with actionable insights to identify ambiguous regions, thereby enhancing diagnostic precision and facilitating more accurate clinical segmentation outcomes. Our code is available at: https://github.fangnengwuyou/DPGNet.

Placenta accreta spectrum (PAS) is a serious perinatal complication. Accurate preoperative identification of patients at high risk for adverse clinical outcomes is essential for developing personalized treatment strategies. This study aimed to develop and validate a high-performance, interpretable machine learning model that integrates MRI morphological indicators and clinical features to predict adverse outcomes in PAS, and to build an online prediction tool to enhance its clinical applicability. This retrospective study included 125 clinically confirmed PAS patients from two centers, categorized into high-risk (intraoperative blood loss over 1500 mL or requiring hysterectomy) and low-risk groups. Data from Center 1 were used for model development, and data from Center 2 served as the external validation set. Five MRI morphological indicators and six clinical features were extracted as model inputs. Three machine learning classifiers-AdaBoost, TabPFN, and CatBoost-were trained and evaluated on both internal testing and external validation cohorts. SHAP analysis was used to interpret model decision-making, and the optimal model was deployed via a Streamlit-based web platform. The CatBoost model achieved the best performance, with AUROCs of 0.90 (95% CI: 0.73-0.99) and 0.84 (95% CI: 0.70-0.97) in the internal testing and external validation sets, respectively. Calibration curves indicated strong agreement between predicted and actual risks. SHAP analysis revealed that "Cervical canal length" and "Gestational age" contributed negatively to high-risk predictions, while "Prior C-sections number", "Placental abnormal vasculature area", and Parturition were positively associated. The final online tool allows real-time risk prediction and visualization of individualized force plots and is freely accessible to clinicians and patients. This study successfully developed an interpretable and practical machine learning model for predicting adverse clinical outcomes in PAS. The accompanying online tool may support clinical decision-making and improve individualized management for PAS patients.

Patients undergoing a mechanical thrombectomy procedure usually have a multi-detector CT (MDCT) scan before and after the intervention. The image quality of the flat panel detector CT (FDCT) present in the intervention room is generally much lower than that of a MDCT due to significant artifacts. However, using only FDCT images could improve patient management as the patient would not need to be moved to the MDCT room. Several studies have evaluated the potential use of FDCT imaging alone and the time that could be saved by acquiring the images before and/or after the intervention only with the FDCT. This study proposes using a denoising diffusion probabilistic model (DDPM) to improve the image quality of FDCT scans, making them comparable to MDCT scans. Clinicans evaluated FDCT, MDCT, and our model's predictions for diagnostic purposes using a questionnaire. The DDPM eliminated most artifacts and improved anatomical visibility without reducing bleeding detection, provided that the input FDCT image quality is not too low. Our code can be found on github.

BackgroundFederated learning (FL) could boost deep learning in neuroimaging but is rarely deployed in a real-world scenario, where its true potential lies. Here, we propose FLightcase, a new FL toolbox tailored for brain research. We tested FLightcase on a real-world FL network to predict the cognitive status of patients with multiple sclerosis (MS) from brain magnetic resonance imaging (MRI). MethodsWe first trained a DenseNet neural network to predict age from T1-weighted brain MRI on three open-source datasets, IXI (586 images), SALD (491 images) and CamCAN (653 images). These were distributed across the three centres in our FL network, Brussels (BE), Greifswald (DE) and Prague (CZ). We benchmarked this federated model with a centralised version. The best-performing brain age model was then fine-tuned to predict performance on the Symbol Digit Modalities Test (SDMT) of patients with MS (Brussels: 96 images, Greifswald: 756 images, Prague: 2424 images). Shallow transfer learning (TL) was compared with deep transfer learning, updating weights in the last layer or the entire network respectively. ResultsCentralised training outperformed federated training, predicting age with a mean absolute error (MAE) of 6.00 versus 9.02. Federated training yielded a Pearson correlation (all p < .001) between true and predicted age of .78 (IXI, Brussels), .78 (SALD, Greifswald) and .86 (CamCAN, Prague). Fine-tuning of the centralised model to SDMT was most successful with a deep TL paradigm (MAE = 9.12) compared to shallow TL (MAE = 14.08), and respectively on Brussels, Greifswald and Prague predicted SDMT with an MAE of 11.50, 9.64 and 8.86, and a Pearson correlation between true and predicted SDMT of .10 (p = .668), .42 (p < .001) and .51 (p < .001). ConclusionReal-world federated learning using FLightcase is feasible for neuroimaging research in MS, enabling access to a large MS imaging database without sharing this data. The federated SDMT-decoding model is promising and could be improved in the future by adopting FL algorithms that address the non-IID data issue and consider other imaging modalities. We hope our detailed real-world experiments and open-source distribution of FLightcase will prompt researchers to move beyond simulated FL environments.

Achieving robust performance and fairness across diverse patient populations remains a challenge in developing clinically deployable deep learning models for diagnostic imaging. Synthetic data generation has emerged as a promising strategy to address limitations in dataset scale and diversity. We introduce RoentGen-v2, a text-to-image diffusion model for chest radiographs that enables fine-grained control over both radiographic findings and patient demographic attributes, including sex, age, and race/ethnicity. RoentGen-v2 is the first model to generate clinically plausible images with demographic conditioning, facilitating the creation of a large, demographically balanced synthetic dataset comprising over 565,000 images. We use this large synthetic dataset to evaluate optimal training pipelines for downstream disease classification models. In contrast to prior work that combines real and synthetic data naively, we propose an improved training strategy that leverages synthetic data for supervised pretraining, followed by fine-tuning on real data. Through extensive evaluation on over 137,000 chest radiographs from five institutions, we demonstrate that synthetic pretraining consistently improves model performance, generalization to out-of-distribution settings, and fairness across demographic subgroups. Across datasets, synthetic pretraining led to a 6.5% accuracy increase in the performance of downstream classification models, compared to a modest 2.7% increase when naively combining real and synthetic data. We observe this performance improvement simultaneously with the reduction of the underdiagnosis fairness gap by 19.3%. These results highlight the potential of synthetic imaging to advance equitable and generalizable medical deep learning under real-world data constraints. We open source our code, trained models, and synthetic dataset at https://github.com/StanfordMIMI/RoentGen-v2 .

Patients undergoing a mechanical thrombectomy procedure usually have a multi-detector CT (MDCT) scan before and after the intervention. The image quality of the flat panel detector CT (FDCT) present in the intervention room is generally much lower than that of a MDCT due to significant artifacts. However, using only FDCT images could improve patient management as the patient would not need to be moved to the MDCT room. Several studies have evaluated the potential use of FDCT imaging alone and the time that could be saved by acquiring the images before and/or after the intervention only with the FDCT. This study proposes using a denoising diffusion probabilistic model (DDPM) to improve the image quality of FDCT scans, making them comparable to MDCT scans. Clinicans evaluated FDCT, MDCT, and our model's predictions for diagnostic purposes using a questionnaire. The DDPM eliminated most artifacts and improved anatomical visibility without reducing bleeding detection, provided that the input FDCT image quality is not too low. Our code can be found on github.

While imaging-genetics holds great promise for unraveling the complex interplay between brain structure and genetic variation in neurological disorders, traditional methods are limited to simplistic linear models or to black-box techniques that lack interpretability. In this paper, we present NeuroPathX, an explainable deep learning framework that uses an early fusion strategy powered by cross-attention mechanisms to capture meaningful interactions between structural variations in the brain derived from MRI and established biological pathways derived from genetics data. To enhance interpretability and robustness, we introduce two loss functions over the attention matrix - a sparsity loss that focuses on the most salient interactions and a pathway similarity loss that enforces consistent representations across the cohort. We validate NeuroPathX on both autism spectrum disorder and Alzheimer's disease. Our results demonstrate that NeuroPathX outperforms competing baseline approaches and reveals biologically plausible associations linked to the disorder. These findings underscore the potential of NeuroPathX to advance our understanding of complex brain disorders. Code is available at https://github.com/jueqiw/NeuroPathX .